B2fxxx

Chinese Government on AI Ethics

2026-04-08T18:31:00.001+01:00

The following is a Google translation of a Chinese government document on AI ethics. It's very clumsy and difficult to follow in places but nevertheless an interesting window into the stated official position of China on AI ethics. If someone has a more accurate translation you can point me to, I'd be very interested in reading it.

Notice of the Ministry of Industry and Information Technology and other ten departments on the issuance of the "Artificial Intelligence Technology Ethics Review and Service Measures (Trial)"

Released 2 April 2026.

Machine translated by Google.

Chapter 1 General Provisions

Article 1 aims to regulate the ethical governance of artificial intelligence technology activities and promote fairness.

Fair, harmonious, safe and responsible innovation will drive the healthy development of the artificial intelligence industry.

Opinions on the Governance of Science and Technology Ethics and Measures for the Review of Science and Technology Ethics (Trial) (hereinafter referred to as the "Opinions")

This regulation is formulated in accordance with laws, regulations, and relevant provisions, including the "Ethics Measures".

Article 2. The artificial intelligence technology activities to which these Measures apply are within the scope of the People's Republic of China.

Activities carried out within the territory of the People's Republic of China that may infringe upon human dignity, public order, and the health of life

The technological ethics risks and challenges brought about by health, ecological environment, and sustainable development

Activities related to artificial intelligence scientific research and technological development, as well as activities conducted in accordance with laws and administrative regulations.

Other artificial intelligence technology ethics reviews required by laws, regulations and relevant national provisions

Science and technology activities.

Article 3. Scientific and technological ethics requirements should be consistently applied in carrying out artificial intelligence technology activities.

Throughout the entire process, we adhere to the principles of enhancing human well-being, respecting the right to life, and upholding fairness and justice.

Reasonably control risks, maintain openness and transparency, protect privacy and security, and ensure controllability.

Adhering to the ethical principles of artificial intelligence technology, and complying with my country's Constitution, laws and regulations.

Relevant regulations.

Chapter Two: Service and Promotion

Article 4. Establish and improve the ethical standards system for artificial intelligence technology, and promote...

Develop relevant international standards, national standards, industry standards, and group standards, and support

Establish an international platform for standardization exchange and cooperation.

Encourage universities, research institutions, medical and health institutions, enterprises and science and technology

Social groups and other entities participate in the formulation, verification, and promotion of ethical standards for artificial intelligence technology.

wide.

Article 5. Promote the construction of an artificial intelligence technology ethics service system and strengthen human rights protection.

Services include monitoring, early warning, testing, evaluation, certification, and consulting for ethical risks in artificial intelligence technology.

Service supply, enhancing enterprises' ability to prevent ethical risks in technology research and development and artificial intelligence.

Efforts will be intensified to support and serve the ethical review of artificial intelligence technologies for small and medium-sized enterprises.

We will increase our efforts to promote international exchange and cooperation on artificial intelligence technology ethics.

Article 6 encourages universities, research institutions, medical and health institutions, and enterprises to participate in this endeavor.

Industry and technology-related social organizations should conduct research on the ethical review of artificial intelligence technology.

We will uphold technological innovation in the ethical review of artificial intelligence and strengthen the use of technology to prevent human-to-human transmission.

Ethical risks of artificial intelligence technology; promoting high-quality data on ethical review of artificial intelligence technology.

According to the orderly open source and open source approach, we will strengthen the development of general-purpose risk management, assessment and auditing tools.

Explore and evaluate technology ethics risks based on application scenarios; promote technologies that conform to...

AI products and services related to technological ethics, protecting the knowledge of technological ethics review technologies.

Identify intellectual property rights.

Article 7. Conduct publicity and education on artificial intelligence science and technology ethics, and leverage the role of science and technology...

The role of social organizations in promoting and educating the public about the ethics of artificial intelligence technology, and encouraging public participation.

Public participation promotes practical demonstrations and enhances public ethical awareness and literacy. It guides the public...

The media should conduct targeted publicity and education on the ethics of artificial intelligence technology.

Article 8 supports universities, research institutions, medical and health institutions, and enterprises.

Industry and technology-related social organizations should conduct education on artificial intelligence technology ethics and related topics.

Training promotes the development of vocational and curriculum systems, and cultivates talent through various means.

Talents in artificial intelligence technology ethics should be cultivated, and talent exchange should be promoted.

Chapter Three Implementing Entities

Article 9 Universities and research institutions engaged in artificial intelligence technology activities

Medical and health institutions, enterprises, etc. are responsible for the ethical review and management of artificial intelligence technology within their respective organizations.

The responsible party should, in accordance with the requirements of Article 4 of the "Ethics Measures", establish a person...

Artificial Intelligence Technology Ethics Committee (hereinafter referred to as the Committee). The Committee should be equipped with...

Necessary staff, office space, and funding, and effective measures should be taken to ensure

The committee operates independently. Qualified organizations are encouraged to conduct research on artificial intelligence.

Certification related to the energy technology ethics management system.

Article 10 The committee's charter, composition, and the duties and obligations of its members shall be in accordance with the "Lun".

The procedures outlined in Articles 5 to 8 of the "Management Measures" shall be followed. The committee should include members specializing in artificial intelligence.

Experts with relevant professional backgrounds in technology, application, ethics, law, etc.

Article 11 Local authorities and relevant competent departments may, in light of actual circumstances, rely on relevant...

Relevant departments shall establish a professional artificial intelligence technology ethics review and service center (hereinafter referred to as the "center").

(Referred to as the Service Center). The Service Center accepts commissions from other organizations to provide artificial intelligence services.

We provide services such as ethical review, verification, training, and consultation for science and technology activities. Service Center

Review and verification services may not be provided simultaneously for the same artificial intelligence technology activity.

The service center should establish standardized management systems and procedures, and be equipped with artificial intelligence technology.

Dedicated personnel with expertise in technical ethics review and service capabilities, subject to review by local or relevant authorities.

Door supervision.

Chapter Four Work Procedures

Section 1 Application and Acceptance

Article 12 addresses artificial intelligence technologies falling within the scope listed in Article 2 of these Measures.

For activities involving artificial intelligence technology, the person in charge should apply to the committee of their unit.

If an organization has not established a committee or the committee is unable to perform its duties in reviewing scientific and technological ethics, it must...

Please apply to the service center entrusted by this unit to conduct a science and technology ethics review. (No specific application is required.)

Individuals should entrust qualified service centers to conduct science and technology ethics reviews.

The person in charge of smart technology activities should submit to the committee or service center in accordance with regulations.

Application materials. The application materials mainly include:

(a) Artificial intelligence science and technology activity plan, including research background, objectives and...

The plan includes the legal qualification documents, personnel information, and funding of the relevant institutions involved.

Source, proposed algorithm mechanism, data source and acquisition method, testing and evaluation

The estimation method, the proposed software and hardware products, the expected application areas and target users, etc.

(II) Assessment and prevention of ethical risks in artificial intelligence technology activities

and emergency response plans, including the potential scientific and technological advancements brought about by the anticipated applications of artificial intelligence technology.

Technology ethics risk assessment, monitoring and early warning measures for technology ethics risks, and potential risks.

Plans to prevent and control risks to scientific and technological ethics, etc.;

(iii) Commitment to complying with the requirements of artificial intelligence technology ethics and research integrity.

Book.

Article 13 The committee or service center shall decide whether to accept the application based on the application materials.

The application will be accepted and the applicant will be notified. If the application is accepted, it should be based on the principles of scientific and technological ethics.

The likelihood and extent of occurrence, emergency situations, etc., are clearly applicable to general, simple, or...

Emergency procedures, depending on the requirements of different procedures, may be carried out in the form of offline or online methods.

For technical ethics review, if the materials are incomplete, the applicant should be fully informed of the required supplementary materials in one go.

Fee.

Section 2 General Procedures and Simplified Procedures

Article 14 The Artificial Intelligence Technology Ethics Review Meeting shall be chaired by the Committee Chairman.

The meeting shall be chaired by a member or their designated vice-chairman, and at least five members shall be present.

This should include the different categories of committee members listed in Article 10 of these Measures. The service center may refer to...

The examination committee shall organize and implement the work in accordance with its regulations.

Depending on the needs of the review, individuals with no direct interest in the relevant field may be invited.

Advisors and experts provide advice. Advisors and experts do not participate in voting at the meeting.

Article 15. The committee or service center shall conduct ethical reviews of artificial intelligence technology.

The investigation should focus on the following aspects:

(a) In terms of human well-being, do artificial intelligence technology activities have scientific and technological value?

Academic and social value; research objectives contribute to enhancing human well-being and achieving social sustainability.

Does it have a positive effect on sustainable development? What are the risks of artificial intelligence technology activities?

Whether or not the benefit is reasonable.

(ii) Regarding fairness and impartiality, the selection criteria for training data, algorithms,

Are the models and systems designed reasonably? Have measures been taken to prevent bias and discrimination?

Algorithmic optimization ensures objectivity in resource allocation, opportunity acquisition, and decision-making processes.

Inclusivity.

(iii) In terms of controllability and reliability, can the robustness of the model and system be guaranteed?

Strength to cope with open environments, extreme situations, and disruptive factors; whether it can

Ensure users can control, guide, and intervene in the basic operation of the model and system; whether

Develop a continuous monitoring plan and an emergency response plan.

(iv) Regarding transparency and explainability, whether the algorithms, models, and... have been reasonably disclosed.

Information such as the system's purpose, operating logic, interaction methods, and potential risks;

Whether effective technical means are adopted to improve the interpretability of algorithms, models, and systems.

(v) Regarding traceability of responsibility, are there measures such as log management in place?

Chu records comprehensive information on all aspects of the data, algorithms, models, and systems to ensure...

The entire supply chain is traceable and manageable; the qualifications of the technical personnel meet the relevant requirements.

(vi) Regarding privacy protection, the collection, storage, processing, and use of data

Whether sufficient measures have been taken for data processing activities and research and development of new data technologies, etc.

Measures are in place to ensure that privacy data is effectively protected.

Article 16 The committee or service center shall, within 30 days of accepting the application,

Make decisions such as approval, requiring revisions and further review, or disapproval. This is necessary in complex situations or when further review is needed.

In special circumstances such as supplementing or correcting materials, the time limit may be appropriately extended and the extension period shall be clearly defined.

For items requiring modification or not approved, the committee or service center should submit a proposal.

Suggestions for amendments or explanations of reasons. If the applicant has any objections, they should submit them within [timeframe] of the decision's service.

You may submit your appeal to the committee or service center within 3 business days of the date of appeal. Grounds for appeal.

If sufficient, the committee or service center should make a new decision within 7 working days.

Article 17. Those in charge of artificial intelligence technology activities shall promptly identify technological ethics.

Manage changes in risk and report relevant changes to the committee or service center.

The committee or service center shall review and approve applications in accordance with Article 19 of the Ethics Measures.

We will conduct follow-up reviews of AI technology activities to promptly grasp the ethical risks of technology.

In response to changes in risk levels, decisions may be made to suspend or terminate related scientific and technological activities if necessary.

The interval for follow-up reviews is generally no more than 12 months.

Article 18. Multiple entities collaborating on artificial intelligence technology activities may, based on…

Based on the actual situation, mutual recognition of the results of ethical reviews of artificial intelligence technologies among different organizations will be carried out.

Article 19. The simplified procedure may be applied in any of the following circumstances:

(a) The likelihood and extent of ethical risks in artificial intelligence technology activities

No higher than the usual risks encountered in daily life;

(ii) Minor modifications to the approved artificial intelligence science and technology activity plan and

Without increasing the risk-benefit ratio;

(III) Follow-up review of artificial intelligence technology activities that have not undergone major adjustments in the early stage check.

Article 20 The committee or service center shall formulate a review procedure applicable to simplified procedures.

The work procedures and tracking frequency. The simplified procedure review is directed by the committee chairperson.

Two or more committee members should be appointed to undertake this task. The service center can refer to the committee's regulations.

The organization will determine the implementation of the work.

During the simplified procedure review process, if the review result is a negative opinion, the review...

In cases where there are doubts about the content of the investigation or disagreements among committee members, the application of Article 1 should be adjusted.

General procedure.

Section 3 Expert Review Procedure

Article 21 The Ministry of Industry and Information Technology and the Ministry of Science and Technology, in conjunction with relevant departments, formulate...

The "List of Artificial Intelligence Science and Technology Activities Requiring Expert Review of Science and Technology Ethics" will be released.

The "Review List" (hereinafter referred to as the "Review List") is dynamically adjusted according to work needs.

Article 22. Conduct artificial intelligence technology activities included in the review list.

After preliminary review by the committee or service center, the organization shall apply to conduct specialized training.

Home review. If multiple units are involved, the lead unit is responsible for the application. Central enterprises,

Universities, research institutions, and medical and health institutions directly under the central and state organs

Report directly to the relevant competent authority to organize expert review. Other units should report to the local authorities.

The relevant authorities will conduct an expert review.

Article 23. Organizations undertaking artificial intelligence technology activities shall comply with the "Ethics"

Article 27 of the Measures stipulates that materials for expert review must be submitted.

Local authorities or relevant competent departments should act in accordance with Articles 28 to 29 of the "Ethical Measures".

Article 30 stipulates that a review expert group shall be established to verify the compliance of the preliminary review opinions.

The review will be conducted on the aspects of sex, rationality, etc., and the review application will be submitted to the relevant authorities within 30 days of receipt.

The applicant organization provided feedback on the review.

Local authorities or relevant competent departments may entrust the service center to carry out the review work.

do.

Article 24 The committee or service center shall make decisions based on the expert review opinions.

The decision to conduct a science and technology ethics review was issued.

Article 25. The committee or service center shall review the personnel included in the review list.

Smart technology activities will be subject to enhanced tracking and review, with review intervals generally not exceeding 6 months.

moon.

If there are significant changes in the ethical risks associated with science and technology, the relevant provisions of Article [number] of the "Ethical Measures" shall apply.

The twenty provisions stipulate that relevant scientific and technological ethics reviews should be conducted again and expert review should be requested.

Article 26. Artificial intelligence technology activities in areas such as deep synthesis, algorithm recommendation,

Registration, filing, and administrative approval are implemented for generative artificial intelligence service management and other aspects.

Regulatory measures will be implemented, and compliance with scientific and technological ethics requirements will be used as a condition for approval and a subject of regulation.

If so, expert review will no longer be required.

Section 4 Emergency Procedures

Article 27 The committee or service center shall formulate artificial intelligence technology ethics

Establish an emergency review system to clarify emergency review procedures during public emergencies and other urgent situations.

Review the procedures and standard operating procedures. Emergency reviews are typically completed within 72 hours.

For artificial intelligence technology activities subject to expert review procedures, prior to expert review...

The review is usually completed within 36 hours.

Article 28 The committee or service center shall ensure emergency review of science and technology ethics.

Ensure the quality and timeliness of the investigation, and strengthen follow-up work and process supervision. When necessary, invite [relevant personnel].

Consultants and experts in relevant fields are invited to attend the meeting and provide advice.

Chapter Five Supervision and Management

Article 29 The Ministry of Science and Technology is responsible for the overall guidance of national science and technology ethics supervision work.

The Ministry of Industry and Information Technology, together with relevant departments, is responsible for the governance of artificial intelligence technology ethics.

Strengthen the coordination and guidance of emergency ethics review work. Each department shall, in accordance with its responsibilities and authority...

Limited to supervising and managing the ethical review of artificial intelligence technologies within this industry and system.

Each locality is responsible for the ethical review of artificial intelligence technologies within its jurisdiction, in accordance with its duties and authority.

Supervision and management work.

Article 30 The unit shall comply with Articles 43 to 44 of the "Ethics Measures".

The fifteen provisions stipulate that relevant information from the committee and artificial intelligence included in the review list will be considered.

Scientific and technological activities are registered through the National Science and Technology Ethics Management Information Registration Platform, and

The Artificial Intelligence Science Department submitted the previous year's committee work report and was included in the review list.

Reports on the implementation of technical activities and other relevant materials. The service center shall comply with the above regulations.

The Ministry of Science and Technology and relevant authorities are registering information related to the ethics of artificial intelligence technologies.

Information is synchronized and shared.

Article 31 Local authorities, relevant competent departments, and those engaged in artificial intelligence technology

The participating organizations should, in light of their own industry, system, and unit's specific circumstances, [make full use of the information].

Channels that can reflect ethical violations and irregularities in artificial intelligence technology, and in accordance with

The matter will be handled in accordance with relevant regulations.

Article 32. In artificial intelligence technology activities or the conduct of artificial intelligence technology...

Violations of these Measures during ethical work will be dealt with in accordance with the relevant provisions of the People's Republic of China.

The Cybersecurity Law of the People's Republic of China, the Data Security Law of the People's Republic of China, and the Law of the People's Republic of China on the Protection of the Rights and Interests of the People's Republic of China

The Personal Information Protection Law of the People's Republic of China, the Science and Technology Progress Law of the People's Republic of China

Punishment.

Chapter Six Supplementary Provisions

Article 33. Where time limits are specified in these Measures, and are not defined as working days, the provisions regarding such time limits shall apply.

All dates are calendar days.

The term "local" as used in these Measures refers to the provincial people's government that is responsible for the relevant work.

The provincial-level management department responsible for the review and management of scientific and technological ethics in the field of intelligence, "related

"Competent authorities" refers to the relevant competent authorities under the State Council.

Article 34 Local authorities and relevant competent departments may, in accordance with the provisions of these Measures,

Develop or revise artificial intelligence standards for this region, industry, or system based on actual conditions.

The system includes regulations and guidelines for the review and service of science and technology ethics. (Science and technology-related social organizations)

Associations can formulate specific guidelines for ethical review and services related to artificial intelligence technologies in this field.

Standards and guidelines.

Article 35. Relevant competent authorities shall supervise the development of artificial intelligence in their respective industries and systems.

Where there are special provisions for science and technology ethics review and services that conform to the spirit of these Measures, those provisions shall apply.

Provisions. For matters not covered in these provisions, the "Ethics Measures" and relevant laws and regulations shall apply.

The regulations shall be followed.

Article 36 This regulation shall be implemented by the Ministry of Industry and Information Technology in conjunction with relevant departments.

Responsible for explanation.

Article 37 This regulation shall come into effect on the date of its issuance.

Attachment: Artificial Intelligence Technology Activities Requiring Expert Review of Technology Ethics

List

appendix

List of AI technology activities requiring expert review of science
and technology ethics

I. It has a strong influence on human subjective behavior, psychological emotions, and life and health.

The development of human-machine integration systems is affected.

II. Algorithms with the ability to mobilize public opinion and guide social consciousness.

Development of models, applications, and systems.

Third, for scenarios involving safety and personal health risks, highly self-reliant...

Research and development of automated decision-making systems for core capabilities.

This list will be dynamically adjusted according to work needs.

Our algorithmic future – Utopia or Armageddon?

2026-03-20T17:04:00.049+00:00

I did a talk for the Open University's final year, undergraduate, computing and communications project students earlier this week. An edited transcript follows below.

Good evening, everyone.

I am going to be talking about algorithmic decision-making machines we are calling artificial intelligence, even though they are not intelligent, in any meaningful sense of the word.

Some industry leaders are promising artificial general intelligence – AGI – will, magically, stimulate economic growth, rectify climate change and fix pretty much all the world’s problems, including curing cancer, and creating a utopia.

One has said he believes regulations and environmental activists are the tools of the antichrist.

One talks of his pride in killing enemies.

Several warn a future superintelligent AI may wipe out humanity.

Whatever their predictions of the possible effects, all of these characters, hype and doom merchants alike, are basically telling the same story – that AGI is inevitable.

The hype of AI has captured hearts and minds at the highest levels of government and commerce.

Politicians are wrapped in fear of missing out and want AI in everything.

CEOs are beginning to get impatient for a payoff on their investments.

AGI may or may not be inevitable. But there are serious geopolitical, environmental, societal, economic and technological questions about whether we should be pursuing it at all.

Meanwhile not enough attention is being paid to how algorithmic decision-making systems are being deployed in practice and with what consequences – in social welfare, border control, policing, security & intelligence, the military, employment, health and other contexts. Let’s consider some of these and think about whether the inevitability of AGI is the appropriate framing for board rooms and governments to be making decisions about our algorithmic future.

But… Before we get into any of that, a bit of history…

Meet John McCarthy. It was McCarthy who coined the term artificial intelligence when, with Claude Shannon, Marvin Minsky and Nathaniel Rochester, he convened the Dartmouth summer workshop on AI in 1956. McCarthy freely admitted choosing the label ‘artificial intelligence’ as a hook to attract funding. That’s right folks, AI began as a sales pitch. And so it pretty much continues to this day.

On the left is Joseph Weizenbaum who created the first chatbot, ELIZA, in 1966. Named after the female leading character in George Bernard Shaw’s Pygmalion, ELIZA’s Doctor variation was a simple key word matching code that bounced rephrased user statements back at them, typically as questions.

On the right is a sample of the kind of responses ELIZA provided, as noted in the paper Weizenbaum published on the program in the Communications of the ACM journal, in January 1966.

Just designed to enable natural language communication with the computer, Weizenbaum was astonished and then disturbed it elicited emotional responses from users, who assumed ELIZA was conscious and showing interest in them. Famously, on at least one occasion, his secretary asked him to leave the room, so she could converse privately with ELIZA.

This false attribution of human empathy, understanding and consciousness to machines became known as the ELIZA effect.

Weizenbaum considered it delusional thinking and was a critic of the blind faith in and adoption of technology generally. He also believed that using algorithms for tasks needing empathy & human judgment (e.g. therapy or welfare or criminal justice) was “obscene”.

I’m going to fast forward, now, through to the early 2010s. AI, machine learning, deep learning, neural networks had lost the power to dazzle benefactors and had spent quite a chunk of the interim in an “AI winter”. Funds had been hard to come by and the field was marginalised.

December 2012 was the inflection point.

Geoffrey Hinton published a paper with his then students Alex Krizhevsky and Ilya Sutskever, entitled “ImageNet Classification with Deep Convolutional NeuralNetworks”. They used GPUs rather than CPUs to create a deep learning image/pattern recognition system they called AlexNet, after its primary developer Krizshevsky. It spawned an almost instant change from AI winter to apparently unlimited venture capital funds for AI researchers and startup companies.

The underlying deep learning technology had not advanced that much since the 1990s but there came a realisation that the computing power which had become available, along with the bottomless bounty of digital data available via the internet, created a huge increase its possible practical applications.

Hinton, who has been called the godfather of modern AI, auctioned their company, DNN Research, the only asset of which was their deep learning paper. There were four bidders (Google, Microsoft, DeepMind and Baidu), Google winning with an offer of $44 million.

Krizhevsky remains one of the most respected AI scholars. Sutskever left Google in 2017 and became a founder and chief scientist at OpenAI where he oversaw the development of ChatGTP [sic]. Though regretting it shortly afterwards, he was one of the board members who ousted Sam Altman in 2023. He stepped down after Altman was reinstated. He founded Safe Superintelligence Inc (SSI Inc) in 2024 with the aim of developing super intelligent AI safely.

By 2012 unfettered, private sector mass surveillance & profiling, of a scale unthinkable before the turn of the century, plus addictive, attention- grabbing apps and social media algorithms, had been long established as the core business model of the internet. They were and are the profit engines of a concentrated set of giant, mostly US and Chinese, technology corporations like Google, Apple, Facebook/Meta, Amazon, Microsoft, Nvidia, Oracle, Tencent (WeChat etc), Alibaba, ByteDance and others.

Notable mention should also be made of the Taiwan Semiconductor Manufacturing Company Limited (TSMC or Taiwan Semiconductor), the world’s largest chip manufacturer which has an almost global monopoly on advanced AI chips, being a key supplier of Nvidia which in turn supplies the hardware at the heart of most advanced AI systems.

The technology systems built and rolled out by those companies are used, extensively, by states in the whole gamut of government services from law enforcement, health and social welfare to border control, military, security and intelligence.

The most profitable firms in the world either monetise or otherwise exploit data through stalker advertising and profiling and/or provide software and hardware services and infrastructure to the economic & state actors who do.

A report in the past couple of weeks, for example, confirmed that the US Customs & Border Protection agency has been buying location data from adtech and data broker industries to track people. The brutal Immigration & Customs Enforcement agency is also buying location data and is actively soliciting more, as well as more, related, surveillance technology.

Advertising technology – adtech systems invisibly, invasively, relentlessly, secretly, at scale, track activity on every internet connected device.

Algorithms aggregate and classify into profiles all those actions.

That makes it a national security issue, according to the Irish Council for Civil Liberties, because these systems, and the vast ecosystems of companies that run and are tapped into them, track and share everybody’s personal information. Including those at the highest levels of government, intelligence, security and military services.

Cameras, sometimes with integrated face recognition, watch us everywhere, in public and in parts of previously private spaces, such as homes. So, for example, GCHQ ran ascheme called Optic Nerve, which surreptitiously collected webcam images, in one six-month period from 1.8 million Yahoo! account holders. In an effort to comply with the Human Rights Act of 1998, operators were limited to collecting one image from each camera every 5 minutes.

Phone apps & networked computers track location and behaviour.

Thousands of economic and state actors collect and/or buy, process and attempt to derive benefits from the digital panopticon we have permitted them to build.

Systems built to collect data, profile us, target ads, grab & maintain our attention and fuel profits have become systems of disproportionate influence and control.

The quest for super intelligent AI that can outperform humans on every dimension is the latest manifestation of all this. The opportunity for those AI models to chomp on the entire digitised history of human creativity and personal data, generated through interaction with the internet, was too good to miss, now the computing power had become available.

And grab that material for training their giant AI models is exactly what the big AI companies have done, riding roughshod over intellectual property, creative moral rights, privacy and data protection rights, in the AI arms race. Hundreds of billions of dollars have been and are being invested in AI companies, models and the massive, environmentally corrosive data centres that run them, leading to further intensification of the data gathering surveillance.

It’s not just corporations and governments, by the way, who have been collaborating in building our mass surveillance society. We have, enthusiastically, invested in and deployed these devices, gadgets and systems and the frictionless convenience and access to information, entertainment, consumption, retailers and community they offer.

And when some of the companies involved get negative publicity for pushing the boundaries of invasive behaviour, we collude in excusing them; or just shrug disinterestedly, if we notice at all.

In one of the gentler examples of this, in 2024 Microsoft discovered state backed hackers from China, Russia and Iran using Microsoft AI for cyber attacks. Security experts assessed the reports and evidence and concluded that Microsoft must have been engaged in large scale, surreptitious spying on their AI users, in order to be able to detect these hackers.

Defenders of Microsoft suggested it was “not fair” to call the company’s actions “spying.” They were just the good guys and they found the bad guys. Besides the small print in the terms of service say, explicitly, that Microsoft monitors users; and all users agreed to the ToS, so had no justification for complaining.

The rest of us pled indifference or simply didn’t pay any attention.

Our expectations of privacy have dropped enormously.

A bit like what fisheries scientists call a shifting baseline. Basically, radical declines in ecosystems or fish species over long periods of time, caused by intensive overfishing, went unnoticed. The declines were evaluated by experts, who used the state of fishery at the start of their careers as the baselines, thereby missing the longer-term trends.

What seems “normal” or “natural” is whatever we experience, as children or at the start of our careers or in particular contexts or circumstances; and we evaluate everything from that baseline.

Surveillance is easier & more pervasive than it has ever been – commerce & governments expand & consume it voraciously.

Not long ago the ability to track people’s daily movements, by knowing their employer, home address or place of worship, was considered a dangerous power, only available, via a warrant, to specific government agencies. Now pretty much anyone can access this capability.

In the last century practically the only people who provided their fingerprints were suspects detained in police stations. Now kids need to press a fingerprint reader to register or get access to lunch, lockers or books in school; or unlock their phones. Children have been conditioned to find it normal to provide their fingerprints for routine daily tasks.

Sadly my call to arms to teenagers to opt out of their school fingerprint systems, as is their right under section 26(5) of the Protection of Freedoms Act 2012, went unnoticed.

After over two decades of the deployment of biometric collection systems in schools, the UK government finally released some guidelines on their use in July 2022. They are short and not particularly enlightening.

Privacy has been disappearing faster than fish stocks, without us paying attention and the AI industry is supercharging that process.

In November 2022 the release of ChatGTP [sic] [thanks to Cory Doctorow for pointing out the error :-). I checked the recording (only available to the students) and it looks like I repeated it throughout the talk] rocked the world. Queue another stampede of venture capital and big tech investment and other economic actors wanting in at the apparently brightening dawn of a new AI age.

Now the arrangement of these investments is hugely complicated… so a large chunk of the $135 billion, 27% stake Microsoft has in OpenAI, for example, is in Microsoft’s provision of Azure cloud services to train and run OpenAI models.

There is massive cross fertilisation with many of the investors hedging their bets, by carving out a piece of many of the big AI companies; in the hope of being there for the payoff, when/if one of them reaches the AGI pot at the end of the rainbow.

Depending on which list you consult, the biggest companies include Nvidia, Microsoft, Apple, Alphabet (Google), OpenAI, Meta (Facebook), Tesla, Oracle, Anthropic, Amazon, Palantir, IBM, xAI, Anduril Industries and Databricks.

You don’t need to worry about most of these guys on the slide if Armageddon does come to pass. Many of them are building giant, luxury, underground bunker complexes, they and their chosen friends, family and acolytes get to ride it out in. I believe Mr Zuckerberg has chosen Hawaii for a 1,400-acre, $270 million compound. Planning documents suggest it will include a 5,000-square-foot underground shelter, with its own energy and food supplies.

I’ve been talking for a while and haven’t yet even addressed the question of what AI actually is. Partly because it is a fraught and complicated one and partly because it remains the catch all sales pitch that John McCarthy conceived it as. AI appears to be everywhere and in everything these days.

Think about how we’d be able to talk about planes, trains, automobiles, bikes, boats, rockets, even walking, if we didn’t have words for them other than “vehicle”.

There’s a media frenzy about a new vehicle that breaks all previous speed records. It happens to be a rocket but the public don’t know that. They want their car, bike etc to be able to go as fast.

That’s pretty much where we are with AI. Just replace the word “vehicle” in that scenario with “artificial intelligence”.

All software tools are being labelled AI and people, in public and policy debates, talk past each other when discussing it because of the paucity of understanding of what precise kind of technology it is they are talking about.

AI covers a huge range of different technologies, some of them nothing to do with AI, some useful e.g. spell checkers, some revolutionary e.g. image processing and breakthrough tech on protein folding, some harmful e.g. mass surveillance tech.

Some deployed in inappropriate contexts and causing harm, like Elon Musk’s Grok chatbot generating estimated 3 million sexualized images including 23,000 of children, over an 11 day period at the end of 2025 and beginning of this year.

Yet many of these tools have the potential to have positive effects. What matters is how we design and use them, who controls them and for whose benefit.

Basically then, AI is umbrella term for a loosely related range of technologies –

LLMs like ChatGTP [sic] have nothing in common with software social welfare depts use to evaluate if someone may be accused of benefit fraud

But how they work, what used for, by whom, against whom & how they fail – is hugely different

Generative AI, GenAI e.g. Stable Diffusion, Claude, ChatGTP [sic], Dall-E can generate apparently realistic content quickly. Progress is impressive but it is unreliable & prone to misuse and generating misinformation

Predictive AI – deployed by governments in policing, border control, social welfare, the military – is where much of the harm happens. It is widely used & being expanded & sold as a “solution” to problems but does not and, often, cannot work. It is hard to predict the future and AI does NOT change that.

Artificial general intelligence (AGI, not to be confused with GenAI) is the holy grail being chased by the AI industry. This is an advanced, sentient AI that can understand, learn, and apply knowledge across a vast, diverse range of domains and that matches or surpasses human capabilities, in all these areas.

A further extension of AGI is artificial super intelligence (ASI) that will evolve from AGI to surpass human abilities by a significant margin.

AI, at its heart, is a probability machine – it finds patterns in data, re-mixes, regurgitates – predictive text/images/code on steroids. It is NOT intelligent or sentient or empathetic or conscious.

Discriminatory data and algorithms have been and are being hardwired into predictive AI systems. The oppression, discrimination, violence that have blighted the lives of marginalised communities, throughout history, is getting baked into AI systems, perceived, wrongly, to be “intelligent”, neutral, decision-making machines. With this, institutional discrimination and prejudice will become harder than ever to combat.

People, not directly affected, can get an idea of the real-life consequences for victims of institutional aggression, based on rogue IT systems, from stories like the PostOffice Horizon scandal.

Bottom line, we should be less worried about future utopia (even an exclusive billionaires’ members only utopia) or Armageddon and more concerned with the harmful effects of these technologies right now and in the recent past, particularly those under the control of powerful concentrated economic and political forces that can, reasonably objectively, be described as malevolent.

Too often the harms suffered by the marginalised are dismissed, by those with vested interests, as inconvenient hurdles to “progress”.

Let’s take a look at just some of the harms that are been perpetrated by people deploying these technologies in the real world.

I’d like to look at examples from some of the areas on the slide.

Dr Joy Buolamwini downloaded some face recognition software to help with an art project when she was at MIT and discovered it could not register her face, unless she put a white mask on. I’ve put video of her describing this experience in the slide and recommend you watch it when you get the chance.

The first independent study of South Wales Police’s use of face recognition, at a Champions League football match and a six nations rugby game, was published by Cardiff University researchers in 2018. At the football the system made correct identifications in 3% (yes I said three per cent) of cases (94 people out of 2710 scanned) and falsely identified people in 72% of cases (1962 people). By August 2020 the Court of Appeal ruled that the force’s use of live face recognition was unlawful.

Advocates, vendors and every techbro who sells face recognition gadgets will tell you the technology is better now.

Some even claim that the encoded bias has been fixed. (See response to Prof Fry's question at 55m 52s).

It remains racially biased and is considered a significant threat to human rights by many, including Buolamwini. A visceral example was when the head of Iran’s agency for enforcing morality laws announced, in September 2022, that face recognition tech would be used “to identify inappropriate and unusual movements,” including “failure to observe hijab laws.”

Face recognition has become popular with authoritarian regimes as a tool to suppress dissent. It is also very popular with the UK government who in December last year pledged to “ramp up facial recognition” and “better equip the police”.

In her 2023 book, Buolamwini documents the evidence of discrimination hard wired and coded into technology like face recognition. It is discriminatory on multiple dimensions, including race, gender, disability, religion.

AI mirrors the bias and assumptions built into its training data, which, in turn, show the bias and assumptions of the societies that data was produced by and about. It was trained on data shaped by generations of policy decisions, societal norms & attitudes and security doctrines that framed ethnic minority or disabled or migrant or female or gender-nonconforming or Muslim or Jewish or other religious or nomadic identities as inherently inferior, less valued or just plain suspicious. It should not be a surprise it exhibits bias.

This is the brilliant Dr Timnit Gebru. At the AI conference in 2015 where Sam Altman and Elon Musk launched OpenAI, she was one of only 5 black people in a delegation of thousands.

Nascent AI systems were already having an influence that caused her deep concern – deciding credit scores, whether to grant mortgages or access to housing, flagging suspects for police, helping judges to decide sentences or whether to grant bail.

A system called COMPAS (Correctional Offender Management Profiling for Alternative Sanctions) was and is being used by judges and parole officers to make decisions about bail, sentencing and parole.

COMPAS attributed high re-offending risk scores to black defendants and lower scores to white counterparts. A 2016 investigation, by ProPublica, showed COMPAS more than twice as likely to be wrong about black defendants as white ones.

Given COMPAS and other discriminatory AI systems she was studying – predictive policing systems trained on historic data which led police to target already over-policed minority communities; Google and Microsoft’s gender biased translation software which assumed e.g. certain professions like doctor or engineer were male, the Google computer vision algorithm that classified black people as apes – Gebru was furious.

As she put it, “A white tech tycoon, born and raised in South Africa during apartheid, along with an all-white, all-male set of investors and researchers” were launching OpenAI allegedly to, as they claimed, stop AI “taking over the world”.

It was clear to Gebru that the narrative generated, to focus media, political and public attention on imagined future Arnold Schwarzenegger like Terminators destroying the world, was intended to distract from the need to address the serious problems that AI systems were already causing.

Working out why AI systems churn out biased information is difficult. Some experts claim it is impossible to tweak models to fix bias, once embedded, because they are so complex, even their creators don’t understand them. Gebru had an idea on how to do a more effective job at trying filter out bias when the systems are getting built.

She proposed every AI training dataset be documented in detail, just as every component in the electronics industry is. Every data set should be “accompanied with a datasheet that documents its motivation, composition, collection process, recommended uses”, details on how it was created, what was in it, what its purpose was to be, what limitations it might have, make the design transparent, reproducible, independently testable. It didn’t catch on.

Big AI firms don’t do transparency, much though they have occasionally claimed otherwise. OpenAI’s standard excuse is that they don’t want to make the details available to bad guys who might misuse them.

Gebru believes we need regulation that imposes transparency on AI companies; that it should always be clear when we use AI systems that they are machines, not sentient, empathic beings; and that the organisations behind them should be required to document and publicly release training data and model architectures.

In 2018 Gebru joined Margaret Mitchell at Google to co-lead the AI ethics research team. Mitchell was a computational linguistics specialist known and respected for her work on fairness in machine learning. She was frustrated that her repeated internal warnings, of the potential problems the AI systems Google were developing would cause, were falling on deaf ears. Worse, often she would get memos from HR telling her to stop making waves and be more collaborative. By 2020 both Mitchell and Gebru became depressed that they could not get through to management about the risks of LLMs.

In 2021, Gebru, Mitchell and a couple of other brilliant computational linguistics experts, Emily M. Bender and Angelina McMillan-Major, who shared their concerns, wrote a seminal paper, On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?

Bender was particularly concerned that even linguistics and AI specialists, who should know better, because they know LLMs are coded mathematical probability machines, were saying, publicly, that LLMs were displaying understanding and consciousness.

The stochastic parrots paper ran to 14 pages and summarised the evidence that LLMs were amplifying bias, the companies building them were becoming more and more secretive about their details and they were under-representing all languages other than English. Mitchell has herself credited as Shmargaret Shmitchell on the paper to emphasise the story it tells.

Gebru and Mitchell went through all Google’s internal processes to get approval for the paper and additionally sent it to more than 20 reviewers they trusted inside and outside the company, to get comprehensive pre-publication feedback.

About a month after submitting it, Gebru and Mitchell were summoned to a meeting with Google executives and told to retract the paper because it was too negative about LLMs. It should be more positive. They didn’t want it associated with Google and anyway Google’s LLMs were “engineered to avoid” the problems they described.

Long story short, Gebru then got sacked (though Google said she resigned) and Mitchell was also fired a few months later, reportedly, when attempting to gather evidence of the company’s discrimination against Gebru and sharing it with external parties.

Their departures, particularly Gebru’s, caused a scandal. There were more than a few articles in mainstream as well as technology press. Some even wondered whether Google were as committed to equality, diversity, inclusion and fairness as they liked to claim. Especially after sacking their two AI ethics leads in quick succession, just because they had raised concerns about the potential risks of LLMs? It also drew significantly more attention to the paper than it might otherwise have received.

Gebru now runs the Distributed Artificial Intelligence Research Institute (DAIR) which she founded after leaving Google. She was named one of the 10 greatest minds in tech by Spanish newspaper El Pais in December 2025.

Mitchell is a researcher and Chief Ethics Scientist at Hugging Face.

Time is pressing on, so I’m going to fast forward through a few of the following slides, but before I do, English residents should note that the Heath Secretary is a big fan of AI and is proposing a few developments in the NHS that everyone really should be more aware of.

The core takeaway from his 10 year plan is that doctor patient confidentiality is to be a thing of the past.

AI, for example, will record and transcribe and send to a central database, likely brokered by Palantir, every word a patient and their doctor shares.

I highly recommend signing up to medConfidential’s newsletter if you would like to know more about this.

According to the UN, the first known use of what they called “lethal autonomous weapons systems such as the STM Kargu-2”, a Turkish manufactured drone shown in the picture, was in the Libyan civil war early in 2020.

The drones “were programmed to attack targets without requiring data connectivity between the operator and the munition: in effect, a true ‘fire, forget and find’ capability”.

Since 2018, UN Secretary-General António Guterres has maintained that lethal autonomous weapons systems are politically unacceptable and morally repugnant and has recommended these systems be banned, under international law.

In December 2024 the UN General Assembly adopted resolution 79/239, basically saying that international law should apply to the use of AI by the military. Given the absence of respect for international law demonstrated by major powers in the past few years, the resolution, in practice, carries little weight.

A UK MoD policy paper on AI in defence, published in 2022, says that real time human supervision of such systems “may act as an unnecessary and inappropriate constraint on operational performance”.

At the time the US, Russia, India, Australia agreed. Brazil, South Africa, New Zealand, Switzerland, France & Germany wanted bans or restrictions.

In August 2023, the US announced “the Replicator Initiative” to produce large numbers of cheap autonomous weapons. The intent, according to a January 2026 briefing for Congress, was “to deploy uncrewed systems en masse, allowing the U.S. military to disperse combat power over a large number of relatively inexpensive systems.” The first contracts for autonomous boats, drones and anti-drone weapons were issued in May 2024.

A House ofLords AI in Weapons Systems Committee report in December 2023 “recommended that the government proceeds with caution on the development and use of artificial intelligence in weapon systems.”

The 26 page response by the government essentially said they’d think about the recommendations but weren’t prepared to be held back when the big bad guys out there were developing this stuff too.

By 2025 Minister for the Armed Forces, Luke Pollard, declared that “compliance with international humanitarian law is absolutely essential”.

The government’s 2025 Strategic Defence Review, however, stated that “The UK’s competitors are unlikely to adhere to common ethical standards” in developing such weapons. It was silent on whether the UK would do so. You may draw your own conclusions on that.

128 countries at the UN weapons convention have not been able to agree to limit or ban the development of lethal autonomous weapons. They have pledged, instead, to continue and “intensify” discussions.

Two weeks ago, a researcher at Cambridge University’s AI Safety Hub released a paper on what he described as military AI agents. He noted they have six structural failure modes that mean, once set up in autonomous weapon mode, there is little or no meaningful human control of them in live military or battlefield contexts.

However well tested (or not) these AI agents might be under laboratory conditions in research & development labs, deploying them in the wild creates profound challenges for safety, security, reliability, trustworthiness and control. Not to mention threats to life.

New Scientist reported, a week earlier, that a researcher at King’s College London discovered that advanced AI models, in war game simulations, escalated the conflict to nuclear strikes, in 95% of cases.

They never agreed to surrender, no matter how badly they were losing; and the made significant mistakes, leading to escalation and unintended casualties in 86% of cases. The simulations were run with GPT-5.2, Claude Sonnet 4 and Gemini 3 Flash.

Major powers are already using AI in war gaming. We do not know what extent they are incorporating AI decision support into live military decision-making processes.

That leads us onto one of the big AI companies, Anthropic, whose CEO Dario Amodei has stated he believes “deeply in the existential importance of using AI to defend the United States and other democracies, and to defeat… autocratic adversaries.” Going on to say:

“Anthropic has therefore worked proactively to deploy our models to the Department of War and the intelligence community.”

Anthropic’s Claude AI is integrated into Palantir software. Palantir’s “Maven Smart System” is variously described, in press reports, as powered by Anthropic’s Claude, to identify, profile, prioritise and select targets for military strikes, in real time.

Palantir’s UK head, Louis Mosley, has said it enables “faster, more efficient and ultimately more lethal decisions where that’s appropriate”.

Anthropic has two restrictions on the use of their technology –

1. That it should not be used for the mass surveillance of Americans (people in the rest of the world are fair game but not Americans)

2. That it should not be used in autonomous weapons systems

They asked Palantir if their AI was used in the bombing of Venezuela and rendition of the Venezuelan president into US custody. Claude did, apparently, play a key role.

To make a long story short, that has led to a falling out with the Trump administration, in turn causing Anthropic’s contract to be cancelled and the company designated a supply chain risk.

It is first time in history such a designation has been applied to a US business. Anthropic are suing the government, while continuing to fulfil their contractual duties, during the six month phase out period Mr Trump has announced; and Claude is being used, by the US military, in their attacks on Iran.

Though it generated approximately 1,000 targets on the first day of the bombings, we don’t yet know if Claude / Smart Maven was used in targeting the Minab school, killing over 170 people, most of them children. If it was, you may wish to consider whether Anthropic & Palantir are responsible for a war crime.

The US government has offered Anthropic’s contract to OpenAI, which Sam Altman has, patriotically, accepted, agreeing not to impose any woke, unreasonable Anthropic-like restrictions. The Defense/War department’s chief technology officer, Emil Michael, described as a “whoa moment” the realisation that Claude was the only AI model authorized in classified settings.

OpenAI are now inside the US classified tent, as is Elon Musk’s xAI and there are ongoing negotiations to include Google too. Michael said he’s not biased. He wants all of them because he needs redundancy.

Whatever concerns any of us might have about AI in weapons, as Alan Z. Rozenshtein, a law professor at the University of Minnesota, has said, the rules governing military AI should not be set through ad hoc negotiations between government officials and individual companies, with no democratic input, no regulatory constraints, and no framework that survives the next change of government.

Dare I add an addendum to Prof Rozenshtein’s concerns, that neither should they be set by the next change of mind of a volatile resident of the Oval Office.

What is not widely known is that a few weeks prior to this dispute, Anthropic’s safeguards research team lead Mrinank Sharma, left the company on the 9^th of February, citing serious ethical concerns; or as he put it in his leaving letter to colleagues: “throughout my time here, I’ve repeatedly seen how hard it is to let our values truly govern our actions.” He’s got an NDA so cannot be more specific about his reasons. But when asked about where he thought we would be with AI safety next year, he posted a gif of the everything’s fine dog.

Here’s some images of the Anthropic dispute with the Pentagon I asked AI to produce. ChatGTP [sic] didn’t do too badly. Not sure what kind of AI juice DeepAI was drinking.

When OpenAI agreed to steal Anthropic’s War Department lunch, without restrictions, their robotics chief resigned, declaring the same red lines as Anthropic. She said, on Twitter: “AI has an important role in national security. But surveillance of Americans without judicial oversight and lethal autonomy without human authorization are lines that deserved more deliberation than they got.”

Silicon Valley has come a long way in the past decade on its attitudes to doing business with the military. Having been reluctant in the 1990s and early 2000s to engage in such business, at least overtly, despite the industry’s origins in the military industrial complex, the more aggressive big tech moguls are now actively touting for the business of the traditional big defence companies, while labelling the incumbents dinosaurs.

In 2024 Google sacked over 50 people for protesting about the company’s joint involvement, with Amazon Web Services, in the $1.2bn Project Nimbus cloud computing contract with the government of Israel, to be used in finance, healthcare, transportation and education, as well as by the military.

In April 2024 Google CEO, Sundar Pichai, sent a memo to all employees, praising their state-of-the-art AI models and research, their leadership in building and deploying responsible and safe AI, the exciting opportunities they have in AI. He concluded with a sting in the tail – a warning that Google is a business and not a place "to attempt to use the company as a personal platform... or debate politics", the implication being that employee protests would not be tolerated.

Former Google chief, Eric Schmidt, served as chairman of the U.S. Defense Innovation Board from 2016 to 2020 and led the National Security Commission on AI from 2018 to 2021.

In 2025 a UN report, From economy of occupation to economy of genocide, listed 122 companies as profiting from genocide, including Google, Amazon, IBM, Microsoft and Palantir. The author of the report, Francesca Albanese, UN Special Rapporteur on the situation in the Palestinaian territories occupied since 1967, was, subsequently, subjected to US sanctions.

Judges and prosecutors at the International Criminal Court have also been sanctioned by the US, following their indictment of Israel’s prime minister, Benjamin Netanyahu for war crimes and crimes against humanity.

Their finances have been frozen and they can’t use a credit card, book a hotel room or so much as buy a cup of coffee. EU member state governments have the legal power and the duty to protect their own citizens, under something called the EU Blocking Statute, but have chosen not to do so, for fear of how the US might respond. Likewise financial institutions and other economic actors.

I wonder what President Eisenhower, who coined the term would make of the modern day big tech military industrial complex?

https://x.com/AssalRad/status/2035480540498006524/photo/1

Israel’s version of the Palantir/Anthropic “Maven Smart System” is what press reports describe as AI called “Lavender”. Lavender was used to generate kill lists of tens of thousands of suspected members of Hamas. These targets’ homes were then bombed, at night, when they were believed to be asleep with their families.

There was little meaningful human oversight, according to six of the Israeli intelligence officers who operated it. Operators were no more than rubber stamps for the AI output, devoting about 20 seconds to each target before authorizing the bombing.

“Dumb” rather than so-called “smart” “precision” bombs were used for suspected junior militants. These cause greater numbers of casualties. One of the intelligence officers reportedly told a +972 magazine journalist that “You don’t want to waste expensive bombs on unimportant people”.

A policy decision was made that it was acceptable to kill up to 20 innocent civilians as well as the Hamas suspect, in the case of low-ranking targets.

The killing of more than a hundred innocents was authorised on several occasions, when the target was a suspected senior Hamas official. Sometimes homes were bombed when the suspect was not there because the information on their whereabouts was not verified in real time.

This was, essentially a decision-making algorithm, targeting people, their families and neighbours for assassination, embedded in a military institution; where organisational dictats determined that the killing of up to 20, or occasionally 100 or more, innocent people was an acceptable measure, in the process of eliminating suspected members of Hamas.

Tolerances for civilian deaths expanded as a matter of organisational, bureaucratic process. When you have a machine that can identify targets, accurately or not, there will always be demands to ID more targets. And if someone in the chain of command thinks those targets are not being dealt with efficiently enough, the pressure to relax the conditions under which they can be eliminated will ramp up.

International humanitarian law prohibits all of this but has no meaning or force for the people of Gaza.

IHL prohibits any attacks directly targeting civilians or civilian objects. They may be incidentally affected by attacks against lawful targets but only if it is proportional; and the attacker must take all feasible precautionary measures to avoid incidental effects on civilians. This includes any foreseeable knock-on, indirect adverse effects on civilians’ life and health such as exposure to toxic chemicals; and the prohibition of the use of starvation as a weapon of war.

By December 2023, Lavender reportedly identified around 37,000 Palestinians as suspected members of Hamas, operating with an accepted error rate of roughly 10 percent.

It is important to understand that AI systems don’t stand alone.

They are never just the technology.

They are always deployed in sociotechnical, organisational, political, social, economic, environmental and, in the case of military systems, geopolitical contexts, with all of the structural flaws and prejudices that come with people and institutions.

And the failure modes of these machines, and the sociotechnical systems they form part of, multiply exponentially, when they run in the real world. In military, health, border control, policing, criminal justice and welfare organisations – which are, themselves, forms of artificial intelligence – these failure modes can be life altering.

Former Greek finance minister, Yanis Varoufakis, spoke to someone who recently left their job at Palantir. He told Varoufakis the “Gaza event”, as he described it, was “very exciting” for technologists. He said what’s happening in Gaza is terrible BUT it was fantastic for Palantir.

He pointed at Varoufakis’s phone and said it was useless to him, at that moment, because it was lying on the table not being used. It’s only when you move with it and interact with it, that it produces data that Palantir can acquire to train their algorithms.

When you bomb people in highly densely populated areas, they panic and do a lot of things that can be tracked – they run around, try to escape, move and are moved, forcibly, from one place to another, they make calls, send messages, try to find loved ones, rush to hospitals, if there are any, and to other places for help.

All that activity was great for training Palantir’s algorithms. And Palantir, having trained their algorithms, using cloud services from AWS, Oracle etc., could develop AI tools to sell to, say the UK NHS, MoD or police forces, with whom they have a combined collection of contracts worth nearly £700 milllion. So, hospitals, for example, can use the Palantir software for managing personnel inside the hospital during emergencies and crises.

The suffering of the Palestinian people in Gaza continues and is intensifying again. The limited amount of humanitarian aid that was being allowed in has been severely reduced, under cover of Israel’s and the US’s war with Iran. On Sunday evening a large wall collapsed onto displaced people’s tents, killing dozens and there are hundreds more injured or missing under the rubble.

We don’t need to wait for some future imaginary, malevolent AI to wipe us out. People in conflict zones all over the world are experiencing dystopian Armageddon right now.

Iran is in the news. Ukraine gets some attention.

Mostly, we are barely aware of their plight.

And by the way, just briefly getting back to autonomous weapons, developments in drone swarms in Ukraine are bringing us closer to the fictional slaughterbots that the Professor Stuart Russell at UC Berkeley was warning of back in 2017.

So where does all this leave us?

We have $100s of billions invested in the race for artificial general intelligence. So far that has been consolidating the increasing concentration of control of the communications infrastructure, that trains and runs the big AI models, in the hands of a small number of big tech companies - the usual suspects, like Google, Amazon and Microsoft, as well as a few I mentioned earlier.

Politicians and board rooms have bought into the hype that they have to embed AI in everything. Though, interestingly enough, a survey of 6000 CEOs, published within the past few weeks, shows they’re getting impatient that investments in AI have shown little return. More than 80% of them said they saw no impact on employment or productivity.

Little or no attention is being paid to the harm that AI tools have been enabling or to mitigating or eliminating that harm.

Cards on the table, I believe the focus on developing AGI is wrongheaded and it is certainly not inevitable as many would have you believe.

It’s a distraction from the good we could be doing with this technology and from fixing the damage we are already causing with it. Whether it is even possible to develop a machine that is better than everyone at everything is largely irrelevant. The mere quest for it is leaving a trail of harm in its wake, some of which we have covered this evening.

We would be far better off investing even a fraction of those hundreds of billions in smaller, specialised AI models, trained with carefully curated, scientifically sound datasets, focussing on solving specialised problems.

The quest for AGI does benefit:

· Investors & big companies developing and selling technology

· The companies grabbing and laundering personal data and creative work of others at enormous scale

· Big economic actors who control the communications infrastructure of AI

· Those trying to replace government services, in health, welfare, policing, criminal justice, defence, health and others with cheap, automated systems, even when, more often than not, are causing serious pain, suffering and distress

AI will do what we make it capable of doing.

Many of the AI models now in existence are impressively capable, in the hands of people who can draw empirically sound, accurate and enlightening outputs from them. Too often they are used by people who don’t understand them, including by researchers. Basic errors in research papers are shockingly frequent, especially when they are using off the shelf AI tools and are not trained computer scientists.

Reviews have determined the majority of machine learning research carried out in areas like health, social science and politics has been shown to be flawed.

When used in life changing areas like policing or immigration, by institutions who don’t understand them, the consequences can be severe or even deadly. It is actively inhumane to embed these black boxes in immigration, welfare, policing & criminal justice, border control, conflict, military, health or employment contexts, as “normal technology”.

It can help to teach, solve problems, big and small, illuminate, inspire and possibly even fulfil some of the hype preached by its high priests but only to the extent that WE are prepared to develop and use these technologies to those ends.

Otherwise, they will be giant, networked, resource draining, socially and environmentally damaging data centres of wires, chips, lights and black boxes, controlled by an increasingly concentrated group of intensely wealthy and powerful economic actors, for their own ends.

We won’t be dealing with Armageddon or Utopia but a dystopia even the combined imaginations of Orwell, Huxley and Kafka didn’t anticipate.

I’ll leave you with an extract from Joseph Weizenbaum’s 1976 book, Weizenbaum, J. (1976) Computer power and human reason: from judgment to calculation. Freeman.

The question is not whether computers can make decisions in the context of human welfare but whether they should. Since we can’t make computers wise, we should not give them tasks that demand wisdom.

Selected sources and further reading

The AI Now Institute

The Open Rights Group

The Distributed Artificial Intelligence Research Institute (DAIR)

medConfidential

Amnesty International

Human Rights Watch

The Hind Rajab Foundation

B'Tselem

Foxglove

Big Brother Watch

Privacy International

United Nations Human Rights Office of the High Commissioner

Electronic Frontier Foundation

European Digtial Rights (EDRi)

Electronic Privacy Information Center (EPIC)

United Nations Relief and Works Agency for Palestinian Refugees in the Near East (UNRWA)

The AI Con: How to Fight Big Tech's Hype and Create the Future We Want by Emily M. Bender and Alex Hanna

Data Grab: The New Colonialism of Big Tech (and How to Fight Back) by Ulises A. Mejas & Nick Choudry

Why Privacy Matters by Neil Richards

Systems Thinking in the Public Sector by John Seddon

The Palestine Laboratory by Antony Loewenstein

Atlas of AI by Kate Crawford

Native: Race & Class in the Ruins of Empire by Akala

AI Snake Oil by Arvind Arayanan & Sayash Kapoor

Your Face Belongs to Us by Kashmir Hill

IBM and the Holocaust by Edwin Black

The Walls Have Eyes: Surviving Migration in the Age of Artificial Intelligence by Petra Molnar

Race After Technology by Ruha Benjamin

Privacy's Blueprint by Woodrow Hartzog

Supremacy: AI, ChatGPT and the race that will change the world by Parmy Olson

Protecting Children in Armed Conflict by Shaheed Fatima KC

Ghost Work by Mary L Gray & Siddarth Suri

Computer Power and Human Reason by Joseph Weizenbaum

The Demon-Haunted World: Science as a Candle in the Dark by Carl Sagan

Drone Theory by Gregoire Chamayou

The Rise Of Big Data Policing by Andrew Guthrie Fergusson

Enshittification: Why Everything Suddenly Got Worse and What to Do About It by Cory Doctorow

The Age of Extraction: How Tech Firms Conquered the Economy and Threaten Our Future Prosperity by Tim Wu

Underground Empire: How America Weaponized the World Economy by Henry Farrell and Abraham Newman

Anatomy of a Genocide: Report of the Special Rapporeur on the situation of human rights in the Palestinian territories occupied since 1967 by Francesca Albanese

From Economy of Occupation to Economy of Genocide: Report of the Special Rapporeur on the situation of human rights in the Palestinian territories occupied since 1967 by Francesca Albanese

Mind the Gap: Foundation Models and the Covert Proliferation of Military Intelligance, Surveillance and Targetting by Heidy Khlassf, Sarah Myers West & Merridith Whittaker

Joint Legal Opinion for the Open Rights Group - The Use of Artificial Intelligence Tools by Government: A Case Study of the Home Office's Asylum Practice by Robin Allen KC, Dee Masters & Joshua Jackson

The Dawn of the AI Drone by C.J. Chivers, New York Times

Congress - Not the Pentagon or Anthropic - Should Set Military AI Rules by Alan Z Rozenshtein

UK Government announcement of £1.2 billion strategic defence partnership with Palantir

Ministry of Defence: Palantir Contracts - Parliamentary debate, 10 February. 2026, Hansard

Human oversight of autonomous weapons doesn't mean much when the aim is to maximise destruction. Lucy Suchman

Human Machine Autonomies by Lucy Suchman & Jutta Weber

EU General Data Protection Regulations (GDPR)

EU ePrivacy Directive

EU AI Act

EU Digital Omnibus - proposal to amend/circumvent GDPR, E-Privacy Directive and EU AI Act

Joint statement On the Proposal for a Regulation as regards the simplification of the digital
legislative framework (Digital Omnibus) by the European Data Protection Board (EDPB) & The Euporpean Data Protection Supervisor (EDPS)

Europe's Hidden Security Crisis by Johnny Ryan and Wolfi Christl for Irish Council for Civil Liberties

America's Hidden Security Crisis by Johnny Ryan and Wolfi Christl for Irish Council for Civil Liberties

Joint statement of security and privacy scientists and researchers on Age Assurance by 400+ privacy & security experts and scholars, March 2026

Children's Wellbeing and Schools Bill Running List of Amendments, p20-21, 11 December 2025

Automating the Hostile Environment: AI in the asylum decision-making system by The Open Rights Group

American Dragnet: Data-Driven Deportation in the 21st Century by the Center on Privacy & Technology, Georgetown Law

Physiognomic Artificial Intelligence by Luke Stark & Jevon Hutson

Joint letter calling for the EU digital security agenda to promote fundamental rights and support a safe digital ecosystem

Geneva Conventions

UN Convention on Certain Conventional Weapons

Note to MP on petition to block mandatory government digital ID scheme

2025-12-01T16:45:00.003+00:00

At the behest of Big Brother Watch, I have written to my MP asking her to attend the parliamentary debate on the petition to block the introduction of digital identity card scheme.

"Dear Layla,

I am writing, as your constituent, to urge you to attend the December 8th petition debate to represent my opposition to the government’s plans for a mandatory digital ID scheme. As you know, the Liberal Democrats have, historically, opposed national identity card schemes and abolished the previous Labour government national identity card when joining in a coalition government with the Conservative party in 2010.
Almost three million people have signed the petition, making it the fourth largest Parliamentary petition recorded. The scale of the response is not a surprise. Millions of people have well-grounded concerns that a mandatory digital ID scheme will pose a threat to privacy, upending our relationship to the state, and make British residents even more attractive targets for hacking by foreign adversaries and criminals. It is critical that the government hears these concerns at the upcoming debate.
The government initially pitched the digital ID scheme as a means to check work eligibility. But since the scheme’s announcement multiple ministers and members of Parliament have made it clear that the initial roll out of the digital ID will merely be the first step in a wider plan to, in the words of Minister for Intergovernmental Relations, Darren Jones, “shut down the legacy state”. Parliamentary Under-Secretary of State for Children and Families, Josh Macalister, has stated the “first use case for it is around right to work checks… we’re not saying we’re going to boil the ocean in one go because the public would be really sceptical of that, so we’re starting with this issue of right to work checks first but there are loads of other applications for digital ID.” The planned mission creep is not even being hidden any more – they are boasting about it.
There is no guarantee that this or indeed a future government would not make digital ID a requirement to access a range of public and private services – including healthcare, education, childcare, tax payments and accessing age-restricted services. A digital ID system will last far beyond this government, meaning that we risk building the mass surveillance infrastructure for a less rights-respecting future administration.
A sprawling digital ID system represents a serious threat to privacy. A mandatory digital ID scheme will gather and store information about us. Each time an individual uses their digital ID, that use may be recorded in government databases, allowing the vast amounts of information to be amassed, searched, and sorted to offer insights through data analysis and profiling. The public should not be forced to bare their lives to the state just to help it administer itself.
The introduction of a mandatory digital ID scheme would fundamentally change the relationship between the population and the state, shifting power away from individuals and towards the government. We currently operate in a system where we can prove our identity as and when we need to with a variety of methods. Mandatory digital ID would turn that dynamic on its head by creating a society on licence, where we might soon need permission every time we interact with the state.
Mandatory digital ID would put the population’s personal data at unprecedented risk of data breaches by creating a honey pot for hackers and foreign adversaries. In the past year, breaches of the legal aid database and Afghans relocating to the UK have resulted in the personal information of hundreds of thousands of people being leaked. The government’s OneLogin system is also reported to be susceptible to impersonation.
A mandatory digital ID scheme will not solve the problems its proponents claim it will and all the usual excuses for it have been rolled out – right to work checks, stop the boats, smash the gangs, halt benefit fraud, border control and multiple variations on tackling the four horsemen of the infocalypse i.e. child abusers, terrorists, organised crime and narcotics cartels. The government, this time, proposed the scheme as a means to crack down on illegal hiring practices,"tackle illegal migration, make accessing government services easier, and enable wider efficiencies." If we have learned anything from the first quarter of the 21^st century, it is that building and deploying infrastructures of mass surveillance does not solve these problems.
The government has not presented a convincing argument that a mandatory digital ID scheme, which will inevitably cost billions of pounds, will persuade criminals who already break employment and immigration law to change their behaviour. In the end, it will be law-abiding people who suffer the effects of digital ID through privacy intrusions and security risks.
One final point that might be worth making is that the digital ID scheme is being pushed heavily by the Tony Blair Institute for Global Change. This “institute” has received huge sums of money from economic actors who stand to benefit significantly from contracts to build and operate the scheme. It reminds us to think of the five questions of a different Tony, Tony Benn –
What power have you got?
Where did you get it from?
In whose interests do you exercise it?
To whom are you accountable?
How do we get rid of you?
Of particular importance in this context are questions 3 & 4.
The infrastructure of mass surveillance that constitutes the modern internet has been repeatedly exploited, by state, criminal and economic actors, to perpetrate widespread abuse, from basic privacy invasion to, tragically, the targeting of innocents for bombing. Another state run digital ID surveillance system will only expand that capacity for abuse. If you build it, they will come – powerful abusers of all stripes, some even believing themselves to be well-intentioned, convinced of their own righteousness – and once deployed mass surveillance infrastructure is incredibly difficult to combat, control or decommission. Please stand in defence of civil liberties, attend the December 8th debate, and articulate the dangers associated with a mandatory government digital ID scheme.
Kind regards,
Ray"

Update, 4/12'25: I got a response from Liberal Democrat MP, Layla Moran. They agree with the concerns raised and have launched their own No to Labour's Digital ID Cards petition.

"Dear Ray

Thank you for taking the time to write regarding the Government’s proposals for compulsory digital ID. Unfortunately, due to previously arranged commitments, I will be unable to attend the petitions debate on December 8^th. However, please rest assured that I and my Liberal Democrat colleagues share your concerns entirely.

The Liberal Democrats are clear that a mandatory digital ID system would cross a red line. It risks eroding long-held civil liberties while doing little to address the Government’s stated aims of immigration enforcement.

We share concerns that a mandatory digital ID system threatens our right to privacy. Digital tools should be about giving individuals more control over their personal data, not giving the government more control over our lives.

We are also concerned that a mandatory Digital ID system could deepen digital exclusion and disproportionately affect society’s most marginalised- older people, people living in poverty and disabled people who often have limited access to digital devices or low digital literacy.

This scheme is set to cost the taxpayer billions. If the government really wants to restore public trust in the immigration system, it could spend this money on Nightingale processing centres, as the Liberal Democrats have called for, to clear the asylum backlog, and still have lots left over to improve public services.

The British public has consistently rejected mandatory ID proposals over the decades, and the Liberal Democrats are proud to have led the charge on this against Tony Blair’s government in the 2000s. We once again stand ready to oppose Labour’s push for mandatory ID cards.

The Liberal Democrats have launched a petition to show the government the strength of public opposition to these plans. You can add your name here.

Thank you once again for taking the time to write.
Best wishes,

Layla"

AI in education – some critical concerns on balancing innovation & ethics

2025-10-03T13:18:00.002+01:00

Some colleagues, Mike Richards and Marian Petre and I recently had a book chapter published on AI in Education: Balancing Innovation and Ethics, in a book edited by Geoff Baker, Professor of Education and Lucy Caton, Senior Lecturer in Education, at the University of Greater Manchester, AI-Powered Pedagogy and Curriculum Design: Practical Insights for Educators.

The chapter is a fairly standard academic treatise with a core set of ten guidelines for the ethical use of AI in education, though my initial starting point for the piece was something of a light polemic which I had not originally intended sharing more widely.

In the wake of President Trump's recent state visit to the UK, when he parked Silicon Valley's tanks on Downing Street's lawns and got a fawning, acquiescent, servile response from the prime minister, however, I've reconsidered.

I'm not going to get into how the government's obsession with growth is misplaced. Professor Richard Murphy does that far better than I ever could. But Mr Starmer's deference to Mr Trump and his preferred Silicon Valley corporations and billionaires, has significant implications for UK sovereignty, the economy, public services generally and education in particular. So without further ado...

AI in education - some critical concerns on balancing innovation & ethics

Vast quantities of data scraped from the internet, often faulty, fallible, incomplete, collected without authorisation or consent, dirty, biased and otherwise problematic on multiple dimensions, get pumped into AI systems to “train” them. These systems are then used to make claims about the world or make decisions affecting real lives that are often not credible or even verifiable.

Large language models (LLMs), like ChatGTP, make probabilistic predictions about what the next word in a sentence should be, in order to spit out apparently plausible text. But the model, though it is assumed, falsely, to be intelligent, neutral and objective, is just making stuff up, however plausible or useful the outputs appear to be. Professor Emily M. Bender of the University of Washington describes these LLMs as synthetic text extrusion machines.

These technologies have been and are being integrated into core social, political and economic systems and institutions. There has been an exponential expansion in police rollouts of face recognition technology in the UK. Algorithms are being used for predictive policing, to decide whether the suspect in a crime should be released on bail and to determine whether someone’s digital profile is sufficiently terrorist-like that they should have a bomb dropped on their home. They are being used to select who should have access to housing and employment.

They are determining which schools kids get to go to and, in schools, children’s access routes to library books or lunch. They are deployed for automated grading of assessment and to decide whether students are attentive enough in class or whether they are cheating on exams. They are targeting people with disabilities for social welfare case review and wrongly accusing hundreds of thousands of fraud. They are becoming core to borders infrastructure deciding whether refugees should be denied passage or let through.

In short, they are being used for classification, rationing the allocation of limited resources and, all too often, inflicting harm.

They have been proven to be racist and discriminatory, disproportionately inflicting harms on historically marginalised groups.

If we are going to continue to build these tools into the infrastructure of education systems, taking the baseline assumption that education is a public good, we have to be asking, right up front, what are the protective or precautionary guardrails?

What are the checks against the harms that inappropriate or poor deployment of these technologies can cause?

What are the safeguards protecting those these tools are used on from the assumptions of the techbros who profit from them?

What are the protections against systemic and institutional racism and discrimination being hardwired into these systems?

What are the restraints on educational institutions’ systemic target/metric chasing harms that get perpetrated against students and staff. e.g. claims re student analytics being the key to student engagement, success and wellbeing. What are the guardrails, in education, when we deploy surveillance-intensive personalised tuition robo-tutor chatbots, automated assessment grading, exam proctoring software, plagiarism detection, personality, intelligence and attention deficit testing?

On the subject of systemic discrimination, how do we ensure we are not solidifying historic patterns of inequity, into these unquestionable/unchallengeable AI black boxes; biases learned from the vast quantities of data, often faulty, fallible, incomplete, collected without authorisation or consent, dirty and otherwise problematic on multiple dimensions, that we are training them on?

Specifically, we need to be asking:

How is the AI being developed?

By whom?

With what purpose?

How does it work and what data is being used to “train” it?

How and through what sources and with what permissions is that data acquired and monitored/reviewed for e.g. biases?

Where is that AI being used?

In what context?

How reliable, accurate, secure and safe is it?

Who benefits?

Who is harmed?

What other problems or opportunities are being generated?

Are the fundamental rights of students and workers being affected? Which & whose rights and how?

How is the AI being tested and monitored in operation?

How is being integrated into not just education but our society, our environment, our politics, our economies, our laws, our technologies?

What are the effects of this?

Who is accountable for those effects?

How do we mitigate harms, improve or evolve or decommission the AI and the economic/ political/social actors benefiting from those harms, when and where necessary?

Many of these questions are sidelined in public debate and institutional and political decision making. Post-hoc superficial approaches to managing AI systems are proffered or preferred. Institutions can put someone in charge of AI ethics and or hire an ethics consultant to run a training seminar for staff periodically, to tell them everyone is responsible for behaving ethically – box ticked, corporate AI ethics responsibility satisfied. Getting on top of building ethical considerations in from inception is really hard. Ticking a box on a form, having an AI ethics policy which gathers dust on a shelf or in a digital archive and putting someone in charge of AI ethics is easy.

How do industry or researchers clean up data, whether for predictive policing or student analytics? Researchers at the AINow Institute are seriously sceptical about capacity of simplistic tech fixes to address this. Importing historical biased data to design your system is the problem. The data can’t be cleaned with any degree of efficacy. And If, as most of these systems are, it is proprietary data processed by proprietary algorithms – no independent verification of the claimed data cleansing process can be performed. All outsiders can see is a black box and ‘trust us, we’re the experts’ public relations from the vendor.

The fundamental issue is that you cannot deploy structurally, systemically and institutionally racist and otherwise discriminatory systems and then engage in post hoc mitigation of harms with any degree of efficacy. By the time AI is embedded in infrastructure it is too late.

As per long held understanding of systems theorists & practitioners, one of the most critical approaches to implementing complex systems, to enable some degree of success, is to involve affected communities in the decision making about those systems, such as the adoption of AI as core educational infrastructure. The people who are obliged to use and those on whom the systems are used are going to be most cognisant of potential harms. They will also be best placed to understand the practical issues around deployment and whether they should be used at all. In the education context, students and staff are the key stakeholders that need to be involved.

There are a host of other questions to consider.

How do we think about how we represent the world and students’ needs in an educational context? What are the institutional and broader politics and ethics of AI in education? What are digital student analytics, exam proctoring, automated assessment, robot tutors proxies for? Are they legitimate, useful, adequate or harmful proxies and for which stakeholders? AI systems are political, not neutral or objective. They are built by people – disproportionately techbros – so it matters who these people are and what their worldview is and what problems they believe they are attempting to solve. They are not education specialists but they are selling ‘solutions’ to educational challenges. ‘Solutions’ that education leaders grab enthusiastically, as they vie to be seen to be innovative, forward thinking, digital futurists and innovators. Not to mention AI solutions offer promises of cost and efficiency savings.

When AI touches complex social, political and economic area such as education, it is critical that we understand the technology, its deployment and its effects, broadly and deeply. Because we are using it to affect people’s lives. Students must be encouraged and taught to be sceptical of these systems or at least analyse them critically and we need vastly more research to test AI systems independently.

The integration of AI into educational infrastructure must be treated with caution and there may well be an argument for the application of a precautionary principle when considering it. Mechanisms for key stakeholders, in particular students and staff, to participate in the decision-making processes around deployment and the facilitation of time, resources, space and agency to understand, assess, and reject such systems are crucial.

Just one final general thought to conclude which may well make all of the above moot:

Are ethics in AI or efficacious ethical guidelines or controls even remotely conceivable, with a technology built through mass exploitation & abuse of labour, including children, egregious, colonial, exploitative rare metal & other resource intensive mining and manufacturing processes, mass copyright infringement, industrial scale privacy invasion, concentrated control of power & capital enveloping it and the use of which consumes hideous amounts of energy and water?

Submission to ICO call for views on ICO approach to regulating online advertising

2025-09-08T18:34:00.009+01:00

I have summited a response to the Information Commissioner's Office's call for views on their approach to regulating online advertising.

It seems clear that the ICO, having concluded, unequivocally, back in 2019, that the adtech industry was egregiously violating the GDPR, are moving, now, towards finding ways to enable the industry to avoid or evade their data protection obligations with impunity.

The pressure from the UK government to facilitate growth will be a factor here. However the ICO has been underperforming on enforcement for quite some time, as their own annual reports, including the latest, and eminent commentators such as Professor David Erdos of Cambridge University and Baroness Young of Old Scone, the former head of the UK’s Environment Agency, have, unimpeachably, noted. As Prof Erdos says, The ICO's

"strong duty to enforce and to respond to complaints has generally not been reflected in ICO practice and, even more concerningly, its accelerating stance points strongly away from rather than towards any expectation of regular and concrete regulatory action...

Especially post-Brexit, it may be argued that UK regulation as a whole is beset by an serious enforcement gap and that the ICO’s track-record merely reflects this. Even if true, this would in no way demonstrate that such an outcome is acceptable either in the abstract or given the UK GDPR’s very specific expectations...

the ICO’s relative performance cannot be explained by a lack of resourcing as it is likely the single best resourced data protection (and freedom of information) authority in the world with approximately 1,000 members of staff. Rather it has been primarily driven by deeply rooted ICO internal culture which has been fuelled by a lack of effective accountability mechanisms for data subjects and by an Information Commissioner who has publicly set his face against full use of the UK GDPR’s powers by, for example, peremptorily degrading fines in the public sector in June 2022 and, without clear evidence, stating in November 2024 that neither high value nor high volume fines against companies were the best way to achieve impact."

Baroness Young was, if anything, more damning when speaking in the House of Lords in 2023,

"We need a powerful and effective regulator. The ICO’s enforcement and prosecution record has not been sparkling, with low levels of enforcement notices, prosecutions and fines. If, when I was at the Environment Agency, I had had as low a level of those as the Information Commissioner has had, I would think I had gone to sleep somewhere along the line."

The ICO's own recent defence of the lack of enforcement action against the Ministry of Defence, following the serious breach involving the disclosure of personal details of thousands of Afghans who worked with British forces during the UK’s presence in Afghanistan, was weak, at best.

This latest consultation, therefore, which seems to be about enabling the adtech industry to continue to generate oceans of what the Open Rights Group rightly call stalker ads, would appear to be another worrying indication of a longer term trend of the inefficacy and regulatory capture of the ICO.

For the past three decades, we have transformed the greatest communications medium in the history of the planet, the internet, into an invasive, toxic, and in the case of the people of Palestine and other marginalised peoples, oppressive and deadly, mass surveillance machine. We have facilitated the development and deployment of this infrastructure primarily in the interests of economic actors who generate their main revenues through targeted aka stalker advertising.

States have, enthusiastically, supported and exploited and engaged in these developments, in their own interests. Essentially the most powerful ecomomic and political forces have all been pushing in the same direction - more mass surveillance, more mass data collection, storage, analysis and processing, more mass privacy invasion. Hypnotised by the belief that the magic computerised mass surveillance machine can deliver their economic or political dreams, decision makers have paid little substantive attention to the social or democratic consequences.

The most powerful economic actors are thriving, even as the services and technology they offer get worse, so they are happy. Policymakers are in thrall to billionaire techbros building machines they claim can solve the politicians' complex socio-political-economic-environmental problems. The techbros are now claiming we should not even worry about burning the planet, in the quest to build an AI machine that will figure out how to fix the problems of global warming, climate change, environmental destruction.

When the magic tech solutions don't work for the politicians and economic actors they become even more obsessed with making them work - techies must just nerd harder. Besides, as their billionaire techbro mates will constantly remind them, the technology is better now, it will work this time, or sometime in the future, if only you give them the license and enough resources. Innovation and growth will see us through.

My colleague at the Open University, Dr. Syed Mustafa Ali, characterises all this as a combination of racial capitalism and digital colonialism.

I would simply ask -

Is the world more stable than it was 30 years ago?

Have we tackled global warming?

How about racism and xenophobia?

Discrimination on all fronts?

War crimes and genocide?

Crime?

Terrorism?

Poverty?

Equity and equality?

Conflict and mass population displacement?

Social welfare?

Fair and equal access to healthcare, treatment and medicines?

Migration, immigration & border control?

Security and intelligence?

Disinformation & misinformation?

Concentration of wealth and power?

Exploitation and abuse of the under-privileged, on every dimension?

Housing?

Employment?

The obsession of the UK government - economic growth, at seemingly any cost?

Enabling people to live in basic comfort and dignity?

Is there a greater recognition of and respect for fundamental human rights?

What problems has this giant mass privacy invasion machine solved?

How well has it solved them?

What other problems has it caused?

How much has it cost, not just in economic terms but in structural social, political, cultural, environmental, in human, personal, community, local, regional, national and global contexts?

Has it been or is it worth it?

How can we better shape and/or retrofit and/or evolve and/or scrap and replace the architectures of these systems in more socially progressive ways, in the public interest?

Perhaps giving the most blatant and flagrant, systemic and systematic violators of basic data protection rights the ICO's blanket blessing, to continue and expand those practices, might not be the most appropriate course of action?

With all that in mind, and when the appalling notion of chat control is coming perilously close to being endorsed by most EU governments, my response to the ICO consultation copied below, has tended, in places to be direct. Please excuse the formatting - it is copied and pasted from the pdf version of my submission via the ICO webform. It should be noted that parts of my response are also guided by and/or edited from the response of the Open Rights Group to the same call for views.

"Submitted to Our approach to regulating online advertising - Call for views
Submitted on 2025-09-05 21:31:09
Advertising purposes and capabilities
1 Ad delivery and billing
What features within ad delivery and billing are the minimum requirements for a commercially viable advertising model, and why?:
Up front I would remind the ICO of your own adtech & RTB report in 2019 specifying:
"general, systemic concerns around the level of compliance of Real Time Bidding (RTB):
1.Processing of non-special category data is taking place unlawfully at the point of collection due to the perception that legitimate interests can be used
for placing and/or reading a cookie or other technology(rather than obtaining the consent PECR requires).
2.Any processing of special category data is taking place unlawfully as explicit consent is not being collected (and no other condition applies).In general,
processing such data requires more protection as it brings an increased potential for harm to individuals.
3.Even if an argument could be made for reliance on legitimate interests,participants within the ecosystem are unable to demonstrate that they have
properly carried out the legitimate interests tests and implemented appropriate safeguards.
4.There appears to be a lack of understanding of, and potentially compliance with, the DPIA requirements of data protection law more broadly (and
specifically as regards the ICO’s Article 35(4) list). We therefore have little confidence that the risks associated with RTB have been fully assessed and
mitigated.
5.Privacy information provided to individuals lacks clarity whilst also being overly complex. The TCF and Authorized Buyers frameworks are insufficient to
ensure transparency and fair processing of the personal data in question and therefore also insufficient to provide for free and informed consent, with
attendant implications for PECR compliance.
6.The profiles created about individuals are extremely detailed and are repeatedly shared among hundreds of organisations for any one bid request, all
without the individuals’ knowledge.
7.Thousands of organisations are processing billions of bid requests in the UK each week with (at best) inconsistent application of adequate technical and
organisational measures to secure the data in transit and at rest, and with little or no consideration as to the requirements of data protection law about
international transfers of personal data.
8.There are similar inconsistencies about the application of data minimisation and retention controls.
9.Individuals have no guarantees about the security of their personal data within the ecosystem."
Having not addressed these concerns, why does the ICO now wish to find ways to permit the advertising industries to continue evade data protection
law?
2 Ad fraud prevention and detection
What features within ad fraud prevention and detection are the minimum requirements for a commercially viable advertising model, and why?:
How is this question within the remit of the ICO?
3 Brand safety, brand suitability and brand compliance
What features within brand safety, brand suitability and brand compliance are the minimum requirements for a commercially viable advertising model,
and why?:
I realise the ICO is under pressure to legalise some targeted online advertising, without consent, so the advertising industry can continue to generate
revenues but, again, how is this question within the remit of the ICO?.
4 Frequency capping
What features within frequency capping are the minimum requirements for a commercially viable advertising model, and why?:
Yet again, how is this question within the remit of the ICO?.
5 Measurement and attribution
What features within measurement and attribution are the minimum requirements for a commercially viable advertising model, and why?:
Yet again, how is this question within the remit of the ICO?.
6 Targeting
What features within targeting are the minimum requirements for a commercially viable advertising model, and why?:
Why is the ICO so concerned with the commercial viability of advertising models. If I can quote from your own website, "The Information Commissioner is
the UK’s independent regulator for Data Protection and Freedom of Information, with key responsibilities under the Data Protection Act 2018 (DPA) and
Freedom of Information Act 2000 (FOIA)." Your "role is to uphold information rights in the public interest" not working out commercially viable advertising
models. You cover
Data Protection Act
Privacy and Electronic Communications Regulations
Environmental Information Regulations
eIDAS Regulation
NIS Regulations
Freedom of Information Act
General Data Protection Regulation
INSPIRE Regulations
Re-use of Public Sector Information Regulations
Investigatory Powers Act
None of these, as I understand the regulations concerned, have anything to do with "What features within targeting are the minimum requirements for a
commercially viable advertising model, and why?"
RTB and behavioural advertising blatantly operate in breach of data protection regulations where, once consent is given (and often, even when it's not)
adtech intermediaries process, share and re-purpose this data at will. This is illegal under the UK GDPR, which requires data not to be processed beyond
the specific, granular purpose for which consent was given.
The true commercial viability of advertising practices cannot be measured despite the claims of the industry on its efficacy. Prices are distorted and the
market is dominated by giant oligopolies and consequent unfair competition of non-compliant advertising practices. Rather than asking what targeting
features are needed to attain a “commercially viable advertising model” should not the ICO be taking action to enforce data protection law so egregiously
ignored by the industry, to remove illegal advertising from the market, and restore a level playing-field for actual law-abiding businesses.
7 How significant are the changes in ICO regulatory posture towards PECR regulation 6 consent requirements that would be required to
enable delivery of a commercially viable advertising model?
Change needed - Ad delivery and billing:
Change needed - Ad fraud prevention and detection:
Change needed - Brand safety, brand suitability and brand compliance:
Change needed - Frequency capping:
Change needed - Measurement and attribution:
Change needed - Targeting:
No change
Please explain your answer::
Why is the ICO concerned with undermining PECR to "enable delivery of a commercially viable advertising model?"
Regarding the changes on targeting there should be 'No change'.
The ability to target individuals based on personal data is the main enabler of harms, discrimination and predatory practices that plague online
advertising. Targeting based on personal data exposes women to unjust prosecutions for their attempt to exercise reproductive health rights; problem
gamblers to being targeted with gambling ads that are meant to exploit their addiction; anyone to be excluded on the basis of their gender, sexual
preferences, ethnicity or other sensitive characteristics; children and those in a more vulnerable status to be targeted and taken advantage of.
These are not unfortunate outcomes, but a feature of the technology. Behaviour is the only personal data that can be observed and captured by storage
and access technologies. It is never a reliable proxy for an individuals' characteristics, preferences or inner desires, but is a reliable means to identify
addiction, health statuses and other syndromes—all of which are, indeed, recognisable by “typical”, “compulsive” behaviours and clearly discernible
patterns of behaviour.
A system that is inherently bad at guessing your commercial preferences but inherently good at identifying weak spots that can be exploited does, not
surprisingly, serve the purpose of exploiting individuals better than it does serve the purpose of delivering legitimate advertising. Advertising systems that
target individuals on the basis of personal data should NEVER be considered low-risk or exempted from consent requirements. That the ICO is running a
consultation on how to enable such activity would appear directly at odds with your duty to act in the public interest.
This call for views includes the following statement: “We will continue to enforce consent requirements for collecting personal information for ad
targeting and personalisation.”
A RELAXATION of the consent requirement for ad targeting, based on ANY amount of personal data, is clearly outside of the scope of this consultation. It
is important, therefore, that the ICO to honour this statement.
Impacts of our approach
8 How far do you agree that the approach outlined in our call for views can identify commercially viable solutions that can also safeguard
people’s privacy and improve user experience?
Strongly disagree
Please explain your answer::
Your call for views appears specifically designed to undermine people's privacy in order to facilitate the ongoing data protection breaching practices of
the industry.
I strongly disagree that such a consultation can remotely "safeguard people's privacy and improve user experience." On the contrary this call for views
displays all the hallmarks of regulatory capture by industry.
DPA tolerance and facilitation of non-compliant advertising practices prevent a meaningful measurement of the true value and commercial viability of
advertising practices let alone those “that can also safeguard people’s privacy and improve user experience”.
The approach of this call for views turns the relationships between commercial viability and “safeguarding people's privacy and improving user
experience” on its head: it is the duty of the economic actors in the advertising industry to commercialise their services WITHIN the boundaries and IN
COMPLIANCE with the norms that have been established by legislation. The UK GDPR and PECR already require advertising to be done in a manner that
safeguards privacy and our agency. The role of the ICO is to enforce these boundaries, NOT to adapt them to meet the needs of non-compliant
advertising firms or to enable such companies to evade their legal obligations.
In the event of exemptions to cookie consent requirements being adopted, “safeguarding people's privacy” would ultimately depend on the limits and
safeguards in place that underpin those exemptions. The call for views provides some clarifications over what will not be exempted, but does not clarify
what practices are being considered to be covered by those exemptions. Without such details it is impossible to evaluate how the ICO is proposing to
“safeguard people's privacy” while conducting this call for views.
9 Would you anticipate any of the following positive impacts if any of the capabilities referenced were permitted without PECR consent in
circumstances where the ICO considers them to be low risk to people? Please select all that apply:
If other, please specify::
I would anticipate NO positive impacts, in the public interest, of enabling industry to evade or circumvent their legal obligations.
Which positive impacts for which stakeholders, in particular, is the ICO interested in? The interests of citizens and customers are not coincident with the
interests of the ecology of economic actors that make up the advertising industry.
Please provide any evidence on the likely scale of these positive impacts::
I refer you to the Irish Council for Civil Liberties report from 2023, Europe's Hidden Security Crises, on the wider NEGATIVE impacts of RTB which
concludes: RTB is not just a privacy concern but a national security concern:
"Real-Time Bidding (RTB) allows foreign states and non-state actors to obtain compromising sensitive personal data about key European personnel and
leaders.
Key insights:
Our investigation highlights a widespread trade in data about sensitive European personnel and leaders that exposes them to blackmail, hacking and
compromise, and undermines the security of their organisations and institutions.
These data flow from Real-Time Bidding (RTB), an advertising technology that is active on almost all websites and apps. RTB involves the broadcasting of
sensitive data about people using those websites and apps to large numbers of other entities, without security measures to protect the data. This occurs
billions of times a day...
...EU military personnel and political decision makers are targeted using RTB...
...Google and other RTB firms send RTB data about people in the U.S. to Russia and China, where national laws enable security agencies to access the
data. RTB data are also broadcast widely within the EU in a free-for-all, which means that foreign and non-state actors can indirectly obtain them, too.
RTB data often include location data or time-stamps or other identifiers that make it relatively easy for bad actors to link them to specific individuals.
Foreign states and non-state actors can use RTB to spy on target individuals’ financial problems, mental state, and compromising intimate secrets. Even if
target individuals use secure devices, data about them will still flow via RTB from personal devices, their friends, family, and compromising personal
contacts.
In addition, private surveillance companies in foreign countries deploy RTB data for surreptitious surveillance. We reveal “Patternz”, a previously
unreported surveillance tool that uses RTB to profile 5 billion people, including the children of their targets...
Cambridge Analytica style psychological profiling of target individuals’ movements, financial problems, mental health problems and vulnerabilities,
including if they are likely survivors of sexual abuse."
See also ICCL's reports on America's & Australia's hidden security crises from 2023 & 2024 and their 2025 RTB complaint to the FTC.
10 Would you anticipate any of the following negative impacts if any of the capabilities referenced were permitted without PECR consent in
circumstances where the ICO considers them to be low risk to people? Please select all that apply:
Worsened customer experience, Increased risk of privacy harm
If other, please specify::
There will be an increased risk of privacy harm. This privacy harm will be accompanied by what you describe as "worsened customer experience." This
questionnaire gives adtech providers ample freedom to argue in favour of removing consent requirements for a range of purposes, as listed in questions
1-6, questions which themselves seem to be outwith with the duties and responsibilities of the ICO.
The call for views does not provide any proposal whose impact on people’s privacy can be commented upon. Further, the call for views allows industry
players to keep their responses confidential, which could prevent industry submissions in favour of deregulation to be scrutinised publicly.
The “scale of these negative impacts" can only be measured when specific proposals are presented.
The shape and form of this call for views makes it likely that responses of the industry will be over-represented. In turn, the views of those concerned
about significant increased risk of privacy harms, that will undoubtedly arise from providing industry with an ICO license to evade data protection laws,
will be under-represented.
Please provide any evidence on the likely scale of these negative impacts::
I refer you to Cory Doctorow's book, longlisted for the Financial Times and Schroders Business Book of the Year 2025, Enshittification: Why Everything
Suddenly Got Worse and What To Do About It.
Doctorow diagnoses the broader issues with unchecked oligopolistic markets online and the negative impacts already out of control. It is exactly the
opposite of easing the pressure on these industries that Doctorow proposes is the way forward, not the ICO's apparent ideas about giving the advertising
industry more freedoms. From the book description:
"Misogyny, conspiratorialism, surveillance, manipulation, fraud, and AI slop are drowning the internet. For the monopolists who dominate online - X,
TikTok, Amazon, Meta, Apple - this is all part of the playbook. The process is what leading tech critic Cory Doctorow has dubbed 'enshittification'. First, the
platform attracts users with some bait, such as free access; then the activity is monetized, bringing in the business customers and degrading the user
experience; then, once everyone is trapped and competitors eradicated, the platform wrings out all the value and transfers it to their executives and
shareholders.
As a result, online public squares have become places of torment, and online retailers are hellish dumpster fires. The virtual gathering places where we
once imagined the world's problems might be resolved are now a sewer of hatred and abuse - thoroughly enshittified.
Doctorow enumerates the symptoms, lays out the diagnosis, and identifies the best responses to these diseased platforms: the monopolies online must
be shattered. Companies too big to fail or to jail - and much too big to care - must be cut down to size. Only an attack on corporate power will permit
effective regulation and real privacy. Tech unions must protect the workers who should, in turn, defend us against their bosses' sadism and greed."
11 Do you see any challenges in delivering commercially viable advertising if the ICO were to revise its regulatory posture towards regulation
6 PECR requirements for specific advertising purposes?
Unsure / Don't know
Please explain your answer::
"Challenges in delivering commercially viable advertising" should be no part of the concern of the ICO.
Diluting the ICO's "regulatory posture towards regulation 6 PECR" to suit an industry notorious for engaging in systematic and systemic practices
breaching its legal obligations under data protection and privacy regulations would seem the very definition of derogation of duty on the part of the ICO.
Technical safeguards
12 Are you aware of any technical safeguards to reduce data protection and privacy risks of storage and access of information for the
advertising purposes listed above?
Please provide your answer::
Though technical architectures can be configured in privacy enhancing ways, with default settings providing better respect for privacy, there are no sole
technical solutions to reduce data protection and privacy risks on the internet.
Structurally, privacy protection requires a combination of regulations, economic incentives/sanctions, social and technological architecture measures
working in harmony to offer the requisite respect for fundamental rights.
According to Schedule 12(2) of the Data (Use and Access) Act:
(3) [...] the means by which the subscriber or user may signify consent include—
10(a) amending or setting controls on the internet browser which the subscriber or user uses;
(b) using another application or programme.
So, powers to amend exemptions to Regulation 6 PECR, which the ICO is proposing should be amended in favour of industry, could actually be used to
give legal enforceability to technical signals.
Giving legal enforceability to technical signals could allow individuals to express consent for online advertising targeting via browser settings and
communicate them persistently as they browse the Internet. But that is only tinkering at the edges of the need for network architecture, regulatory,
economic incentive and social reforms.
13 Do you currently use any technical safeguards or PETs in your online advertising model?
Please provide your answer::
I don't have an online advertising model.
14 Are you aware of any recent innovations which significantly reduce the data protection and privacy risks of one or more of the capabilities?
Please provide your answer::
No and even if there were, there are no simple tech fixes to complex sociotechnical/economic/tech-architectural systemic problems.
That politicians and media and regulators like to believe there are is a fundamental part of the reason why we keep making the same mistakes in failing
to shape the development and deployment of these technologies in the public interest.
About you and your organisation
15 Are you responding on behalf of an organisation?
I'm not responding on behalf of an organisation
If other please specify::
16 If you are not responding on behalf of an organisation, are you answering as:
An academic
If other, please specify::
Final comments
21 Before completing this call for views, do you have any final comments you have not made elsewhere?
Please provide your comments::
I would just like to repeat my concern that the shape and form of this call for views makes it likely that responses of the adtech industry will be
over-represented. In turn, the views of those concerned about significant increased risk of privacy harms, that will undoubtedly arise from providing
industry with an ICO license to evade data protection laws, will be under-represented.
22 We may wish to contact you for further information on your responses. If you are happy to be contacted, please provide your name and
an email address below.
Please provide your name:
Ray Corrigan

23 We may publish in full the responses received from organisations or a summary of the responses. If so, we would like your permission to
publish your consultation response. Please indicate your publishing preference:
Publish response"

Liberal Democrats perspective on e-Visas

2025-06-12T16:22:00.000+01:00

I've had a response from Layla Moran's office to my email about the government planning to use e-Visas for immigration raids.

"Dear Ray,

Thank you for taking the time to share your concerns about e-Visas.

As Co-President of The European Movement, Layla is deeply concerned by the growing number of cases where EU citizens have faced severe consequences due to errors or failures in the Home Office’s digital system. As you rightly point out, problems with proving immigration status can affect everything from healthcare access and housing to employment and education, all of which are fundamental rights people should be able to rely on.

We have heard of real-life experiences which highlight a system that is failing in both transparency and reliability, and we agree that this must be urgently addressed. The Liberal Democrats agree with Settled that a status system that leaves people in limbo, unable to prove their right to live, work, or access essential services in the UK, is not fit for purpose.

We believe EU citizens deserve a secure, reliable and inclusive form of proof of status. The Home Office must take urgent steps to restore confidence in the system, including commissioning an independent review into the operation of the eVisa platform and ensuring there are accessible backup forms of status available. Everyone who has the legal right to be in the UK should be able to prove it easily and consistently.

We continue to call for stronger UK–EU relations that go beyond the instability and bureaucracy of recent years, including by supporting Early Day Motion 1318. This EDM welcomes the progress made at this week’s UK–EU Summit, but also expresses concern that there is still much more to be done, not least on restoring people’s mobility rights and addressing the red tape many now face.

We greatly appreciate the work that Settled has done and continues to do. Please rest assured that Layla and her Liberal Democrat colleagues will keep pressing the Government to show greater ambition in resetting our relationship with Europe and to ensure that those who live, work, and contribute to communities in the UK, including long-standing EU residents, are treated with the dignity and security they deserve.

Thank you again for getting in touch.

Best wishes,

Office of Layla Moran
Liberal Democrat Member of Parliament for Oxford West & Abingdon "

Ban police use of face recognition technology

2025-05-20T16:25:00.003+01:00

At the suggestion of the Open Rights Group, I've written to my local council about police use of face recognition technology.

"I am writing to urge you to to ban police use of live facial recognition technology in Abingdon East.

These surveillance tools are being deployed by police without public consent, or clear legal grounds, undermining the right to privacy, the right to be presumed innocent and making police misuse of the technology more likely.

These tools do not predict crime – they predict policing. Built on flawed, discriminatory data, they disproportionately target Black and racialised communities, low-income areas, and migrants. Rather than making our communites safer, these technologies reinforce racism and criminalise poverty.

“Predictive policing” strips away our fundamental right to be presumed innocent. It creates fear, not safety. The European Union has already recognised these harms and taken action to ban such technologies. The UK must do the same.

It is well documented that when police are given new tools, it is Black and racialised communities that bear the brunt of the harm, with consequences including increased police harassment, injury from use of force and unjust stop and search. I am deeply concerned for the safety of my neighbours if we are to allow police use of “crime predicting” technology to become anymore embedded in our neighbourhoods.

We all want to feel safe in our local areas, but real safety comes from investing in our communities – not from surveillance that fuels fear and distrust.

That is why I urge you to stand in solidarity with your local community, by bringing forward a motion to ban this dangerous tech. As your constituent, I want to feel confident that my rights and safety are being protected.

Please take urgent action to prohibit the use of “crime-predicting” technologies and protect the rights and freedoms of all."

Lib Dem response on Data Use & Access Bill

2025-05-20T16:19:00.000+01:00

I've had a reponse from the Liberal Democrats on my concerns about the Data Use & Access Bill.

"Dear Ray,

Thank you for taking the time to write. We are responding on Layla’s behalf whilst she takes parental leave. You can view the full transcript of the Bill’s third reading, along with a record of Layla’s proxy votes on amendments, here.

The Liberal Democrats welcome the omission of many of the more objectionable elements of the previous DPDI bill, which was introduced by the previous Conservative government but fell when the General Election was called.

Despite these changes, retention and enhancement of public trust in data use and sharing is a major issue in the bill. The focus on smart data and sharing of government data means that the Government must do more to educate the public about how and where our data is used and what powers individuals have to find out this information.

There are still major changes proposed to the Data Protection Act 2018 (GDPR), such as in regard to police duties and Automated Decision Making, which continue to make retention of data adequacy for the purposes of digital trade with the EU of the utmost priority in considering any changes.

We continue to believe that GDPR is not in need of fundamental reform, but rather, where there is any ambiguity in interpretation, clarifications incorporating relevant recitals to the GDPR should be made in the legislation and in improved guidance.

The Liberal Democrats will continue to follow the progress of this Bill closely. Thank you once again for taking the time to get in touch.

Best wishes,

Office of Layla Moran"

Bottom line - the Lib Dems are not concerned about the expansion in data sharing proposed in the Bill, just that "that the Government must do more to educate the public" about it and the GDPR is not in need of reform.

Government plan to use eVisa scheme for immigration raids

2025-05-14T11:28:00.001+01:00

The government has announced this week that the Home Office's flawed eVisa scheme will be used for immigration raids. So I've followed up my email to my MP to alert her to this development. The Open Rights Group are coordinating an effort to make MPs aware and ask for resistance to this scheme.

Dear Layla,

I wrote to you recently about the Home Office’s flawed eVisa scheme.

This week, little noticed in media reporting on the Prime Minister’s unconscionable ‘island of strangers’ targeting of immigrants speech, the Government announced that the eVisa scheme would be used to support immigration raids. If this goes ahead, people with the legal right to be in the UK could be deported because of flaws in the Home Office’s systems.

Since the rollout of the eVisa scheme, the human rights organisation, the Open Rights Group, has heard about travellers stranded at airports, refugees unable to rent a home or get a job, and even a man being made homeless because of a data error. But these harms would pale into significance if eVisa data is used for immigration raids that result in deportation.

Please will you contact the Secretary of State for the Home Department, Yvette Cooper and urge her to stop eVisa data from being used for immigration raids and press again for the government to provide offline alternative for people to prove their immigration status when the eVisa is not working. The seeds of another Windrush scandal have been sown and are sprouting.

Regards,

Ray

Response from government over concerns about Online Safety Act

2025-04-30T12:13:00.002+01:00

My MP, Layla Moran, has had a response from the Secretary of State for Science, Innovation and Technology, Peter Kyle, regarding concerns about the impact of the Online Safety Act on small, safe and harmless websites, and services.

Short version:

The Act is not intended to go after smalll websites and services. Small websites only have to follow the rules to produce risk assessments, defend those risk assessments if they conclude risks are low, provide complaints processes, police content, keep up to date with Ofcom's associated regulations & "Regulation Checker" & 'Digital Support Service', engage with Ofcom support services and have an Ofcom liason officer; and when it comes to compliance, Ofcom will focus on services where the risk and impact of harm is highest.

Mr Kyle concludes by hoping that this information provides some clarity and reassurance.

From my perspective the short answer is: Nope. The longer answer is... you know I don't currently have the capacity for this, so would refer you to the ever insightful, Graham Smith, generally for enlightening commentary on the Online Safety Act.

MPs propose amendments to Data Use & Access Bill

2025-04-29T09:34:00.003+01:00

I've written to my MP, again, about the Data Use & Access Bill, which is due back for a vote in the House of Commons next week, requesting her support for digital rights amendments to the Bill proposed by Siân Berry of the Greens, Alex Sobel from Labour and Steff Aquaone of the Lib Dems.

"Please pass on my thanks to your colleague, Steff Aquaone MP, for proposing an amendment to the Data Use and Access Bill that creates a right to use non-digital forms of ID, which will, hopefully, improve accessibility and digital inclusion.

I would also ask that you support two other important amendments to the Bill, when it comes back to the House of Commons in early May, tabled by Sian Berry of the Green Party and Alex Sobel of Labour. The Bill is scheduled for its report stage on the 7th May.

These amendments aim to improve the protection of personal data for vulnerable groups — something I hope you’ll agree is essential.

1. Supporting those most at risk of harm
The Information Commissioner’s Office currently faces a serious backlog in handling complaints. This affects everyone, but especially vulnerable individuals. For example, survivors of domestic abuse and gender-based violence often face heightened risks from online stalking and other forms of digital harm.

Siân Berry is proposing an amendment to introduce a statutory complaints procedure for people in vulnerable situations, along with a right of appeal to the Information Tribunal. This would give those most at risk a clearer, fairer route to justice.

2. Protecting data from unsafe overseas transfers
The Bill also makes it easier for personal data to be transferred outside the UK — including to countries that lack strong data protection laws. Once data is moved abroad, it can be accessed or misused in ways that would be illegal in the UK.

For vulnerable groups — such as refugees, survivors of abuse, and people facing persecution — this poses serious risks. A single data breach overseas could have life-altering consequences.

Alex Sobel MP has tabled an amendment to strengthen safeguards for international data transfers. It would help ensure that UK citizens’ data is not exposed to harmful misuse abroad.

3. The right to non-digital ID verification
The bill establishes a framework for digital ID verification services. It is absolutely critical that people should continue to have the right to use a non-digital form of ID. This would improve accessibility and digital inclusion. As you know, Steff Aquarone MP has tabled this amendment.

A copy of these amendments are included below and is also available on the amendment papers for the Bill at

https://publications.parliament.uk/pa/bills/cbill/59-01/0179/amend/data_use_rm_rep_0422.pdf

Thank you for your time, and I look forward to your reply.

Siân Berry MP - Complaints procedure for vulnerable individuals - NC15

Siân Berry
.To move the following Clause—
“Complaints procedure for vulnerable individuals
(1) The Data Protection Act 2018 is amended in accordance with subsections (2)
to (4).
(2) After section 165(3) insert—
“(3A) For complaints under subsection (2), the Information Commissioner
must provide appropriate complaints-handling procedures for—
(a) victims of modern slavery,
(b) victims of domestic abuse,
(c) victims of gender-based violence, or
(d) data subjects otherwise in a position of vulnerability.
(3B) Procedures under subsection (3A) must include—
(a) appropriate support for vulnerable individuals;
(b) provision of specialised officers for sensitive cases;
(c) signposting to support services;
(d) provision of a helpline;
(e) de-escalation protocols.”
(3) After section 166(1)(c) insert—
“(d) fails to investigate a complaint appropriately or take adequate action to remedy findings of inadequacy.”
(4) After section 166(2)(b), insert—
“(c) to use formal powers as appropriate to investigate a complaint and to remedy any findings of inadequacy, unless the request from the data subject is manifestly unfounded or excessive.””

Member's explanatory statement

This new clause would require the Information Commission to introduce a statutory complaints procedure for individuals in a position of vulnerability and new grounds of appeal to an Information Tribunal.

Alex Sobel MP –Data transfers overseas - Amendment 10

Alex Sobel
.Schedule 7, page 201, line 5, at end insert—
“(1B) A third country cannot be considered adequate or capable of providing
appropriate safeguards by any authority where there exists no credible means
to enforce data subject rights or obtain legal remedy.
(1C) For the purposes of paragraph 1A, the Secretary of State must make a
determination as to whether credible means are present in a third country.
(1D) In making a determination regarding credible means, the Secretary of State
must have due regard to the view of the Information Commissioner.
(1E) Credible means do not exist where the Secretary of State considers that any
of the following are true:
(a) judicial protection of persons whose personal data is transferred to that
third country is insufficient;
(b) effective administrative and judicial redress are not present;
(c) effective judicial review mechanisms do not exist; and
(d) there is no statutory right to effective legal remedy for data subjects.”

Member's explanatory statement

The amendment would prohibit personal data transfer to countries where data subject rights cannot be adequately upheld and prohibit private entities from using contracts to give the impression that data security exists.

Steff Aquarone MP - “Right to use non-digital verification services - NC7
(1) This section applies when an organisation—
(a) requires an individual to use a verification service; and
(b) uses a digital verification service for that purpose.
(2) Where it is reasonably practicable for an organisation to offer a non-digital
method of verification, the organisation must—
(a) make a non-digital alternative method of verification available to any
individual required to use a verification service; and
(b) provide information about digital and non-digital methods of
verification to those individuals before verification is required.”

Member's explanatory statement
This new clause would create a duty upon organisations to support digital inclusion by offering non-digital verification services where practicable."

UK Government should allow physical proof of immigration status

2025-04-03T14:04:00.001+01:00

In the face of the failing rollout of the UK government's eVisa scheme, I've written to my MP, again, at the suggestion of the Open Rights Group.

"Of the many failures of the Labour government since coming into power last year, their adoption of Conservative/Reform right-wing rhetoric and policies on immigration has been one of the most egregious.

Please can you take action to help people who are facing serious harms because of the Home Office’s flawed eVisa scheme.

The eVisa scheme was supposed to replace physical documents with online proof of immigration status. But the roll-out of the scheme has been chaotic. Technical and administrative problems mean that many people cannot create or access their eVisa at the crucial times that they need it. Many people are currently having to rely on expired documents to prove their right to be in the UK.

Since the scheme was supposed to come into effect on January 1, 2025 people who have the right to be in the UK have been prevented from boarding planes and had their travel plans disrupted. Refugees have been unable to rent homes or get jobs. At least one person has been made homeless because of an error in their eVisa. There are undoubtedly thousands more people facing such harms – including many of your constituents.

The immediate solution:

The Home Office recently updated its guidelines to allow migrants with legacy paper documents to continue using them. This could be extended to all migrants that have the right to remain in the UK. Having a physical document as a back up will mean that people can prove their immigration status even if they are unable to show their eVisa because of technical glitches or Internet outages. It will also help people who do not have a smartphone or who are not digitally literate.

Please will you contact the Minister for Migration and Citizenship, Seema Malhotra and urge her to take this simple step to allow people to have an optional physical document to prove their right to be in the UK"

Save small sites & services from the Online Safety Act

2025-03-14T10:54:00.000+00:00

At the request of the Open Rights Group I have written to my MP, Layla Moran, about the potential impact of the Online Safety Act on small, safe sites and services.

Dear Layla Moran,

I am a constituent of Oxford West and Abingdon.

I am writing because I am concerned that many small, safe and harmless websites, and services are closing down in the UK, because of the duties, risk assessments and threats of fines and even imprisonment contained in the Online Safety Act.

Even a small personal blog, with user comments, must comply with the new rules.

The Act comes into force on 17 March 2025. The Secretary of State, as I understand it, can change the rules very quickly to exempt small websites, under Section 220 of the Act. They could extend the onboarding period, while working out how to ensure small websites, and services do not have to close down.

Overseas websites are also likely to block access to the UK, if they understand the risks.

This could include many services which are designed as safe spaces for friendly conversation, like many Mastodon social media sites. These serve the LGBTQ+ community, for example, to keep users safe from bullying and hate speech.

Yet they may block the UK because of Online Safety compliance risks.

This will impact me personally as I have maintained a blog, B2fxxx: Random thoughts on law, the internet and society, available at https://b2fxxx.blogspot.com/ for over 20 years. I have written more than 4700 blogposts in that time on a wide range of technology policy issues, including, for example, copies of submissions I have made to UK government, EU and other consultations. The Online Safety Act means I have to consider shutting down the blog and deleting all of those writings.

The Online Safety Act is meant to keep people safe, not to shut down safe websites.

Please make Peter Kyle the Secretary of State for Science, Innovation and Technology aware of these problems and ask for a response.

More information:
https://www.newscientist.com/article/2461213-hundreds-of-small-websites-may-shut-down-due-to-uks-online-safety-act/
https://www.telegraph.co.uk/business/2024/12/17/hundreds-of-websites-to-shut-down-under-chilling-internet/
https://onlinesafetyact.co.uk/in_memoriam/
Yours sincerely,
Ray Corrigan

Response to Ofcom technology notices consultation

2025-03-07T13:46:00.001+00:00

At the behest of the Open Rights Group, I have written to Ofcom regarding their technology notices consultation.

Dear Ofcom consultation team,

I am responding to your consultation as an individual.

I am happy for you to publish this response.

I wish to respond to the following consultation question.”

‘Do you have any views on our audit-based assessment, including our proposed principles, objectives, and the scoring system? Please provide Evidence to support your response.’

It is my view that OFCOM needs to consider how they score and consider the following risks and threats that could arise from accrediting any scanning technologies:

1. The threat that the system might infringe on people’s human right to free expression or privacy. This is particularly relevant given these systems could break end-to-end encryption and the recent ECHR ruling in the case of Podchasov v. Russia – https://hudoc.echr.coe.int/eng/?i=001-230854.

OFCOM will have a legal duty to assess the proportionality of any such system that doesn’t infringe on our human rights and could find itself facing legal challenges over this issue if it doesn’t demonstrate an assessment of the impact on the right to privacy from breaking E2EE within its framework.

2. The risk of false positives & wrongful accusations. If these are too high then law-enforcement agencies will be flooded with false positive results from any scanning system and people will be wrongfully accused causing them harms. A higher threshold should therefore be applied to accuracy.

3. The risks are that any system will break UK data protection laws and/or undermine the nation's cybersecurity by introducing backdoor vulnerabilities to private and secure messaging systems. The recent situation where Apple has withdrawn its advanced data protection product from the UK market highlights that forcing or approving a poor technology upon a company could result in UK users losing access to products. OFCOM should consider the risks to UK consumers of forcing new technologies onto providers that are not feasible to deliver or have too high economic and social costs.

4. Equalities act implications and impact on people with protected characteristics– OFCOM will have to consider the impact of any scanning system in relation to the public sector equalities duty.

5. Higher weighting in framework around risks where there are legal duties. The current minimum threshold for ‘fairness’ does not consider the risk Ofcom faces of breaking its legal obligations to consider Human Rights Act, Equalities Act and the Data Protection Act. As such a separate scoring and risk assessment should be taken for each technology it considers to ensure Ofcom can evidence it has met its statutory legal duties.

6. The risk any system will facilitate the spread of CSEM – A regulator wishing to control the use of image-based sexual abuse (IBSA) removal tools must carefully assess the risks posed by perceptual hash inversion attacks. These attacks could result in someone creating CSEM images from the hashed data the tool was using. For Evidence of these attacks, see S. Hawkes, C. Weinert, T. Almeida and M. Mehrnezhad, "Perceptual Hash Inversion Attacks on Image-Based Sexual Abuse Removal Tools," in IEEE Security & Privacy, doi: 10.1109/MSEC.2024.3485497. Further risks and threats from scanning technologies are set out in 'Bugs in our pockets: the risks of client-side scanning - Harold Abelson, Ross Anderson, Steven M Bellovin, Josh Benaloh, Matt Blaze, Jon Callas, Whitfield Diffie, Susan Landau, Peter G Neumann, Ronald L Rivest, Jeffrey I Schiller, Bruce Schneier, Vanessa Teague, Carmela Troncoso, Bugs in our pockets: the risks of client-side scanning, Journal of Cybersecurity, Volume 10, Issue 1, 2024, tyad020, https://doi.org/10.1093/cybsec/tyad020
perceptual hash inversion attacks. These attacks could result in someone creating CSEM images from the hashed data the tool was using. Evidence of these attacks and risks are published in a research paper in IEEE Security & Privacy ( https://ieeexplore.ieee.org/document/10762793)

Yours sincerely,
Ray Corrigan

In relation to item 3. above I'd also recommend those interested read and thoroughly digest 'Keys under doormats: mandating insecurity by requiring government access to all data and communications' by Harold Abelson, Ross Anderson, Steven M. Bellovin, Josh Benaloh, Matt Blaze, Whitfield Diffie, John Gilmore, Matthew Green, Susan Landau, Peter G. Neumann, Ronald L. Rivest, Jeffrey I. Schiller, Bruce Schneier, Michael A. Specter and Daniel J. Weitzner.

Bottom line there is no backdoor that can be created to a cryptographic system that only the good guys will have access to.

Lib Dem position on Data Use & Access Bill

2025-02-13T14:07:00.002+00:00

I've had a response from the Liberal Democrat MP, Layla Moran in relation to concerns raised about the proposed DUA Bill.

Dear Ray

Thank you for writing to me about the Data (Use and Access) Bill.

We welcome the omission of many of the more objectionable elements of the previous Data Protection and Digital Information Bill, which was introduced by the previous Conservative government but fell when the General Election was called. I spoke in the second reading debate of that Bill and you can read my speech here.

Despite these changes, retention and enhancement of public trust in data use and sharing is a major issue in the bill. The focus on smart data and sharing of government data means that the Government must do more to educate the public about how and where our data is used and what powers individuals have to find out this information.

There are still major changes proposed to the GDPR -e.g. as regards police duties and Automated Decision Making which continue to make retention of data adequacy for the purposes of digital trade with the EU of the utmost priority in considering any changes.

We continue to believe that the GDPR /Data Protection Act 2018 is not in need of fundamental reform but rather where there is any ambiguity/difficulty in interpretation clarifications incorporating relevant recitals to the GDPR should be made in the legislation and in improved guidance.

We are also supportive of Baroness Kidron’s amendments to strengthen the rights of the creative industry to challenge generative AI business practices which seek to train models on their work. This is a key campaigning area for the Liberal Democrats, as we believe existing copyright law should be enforced to protect the UK’s creative industries – a world-leading British export.

The Government-as did the last one- continue to claim huge benefits -nearly £10bn. from reforms in the bill. We are very sceptical about this and will continue to scrutinise how these changes will impact business.

Liberal Democrats believe that the UK should be broadly aligned with the forthcoming EU AI legislation in terms of safeguards. Whilst we agree that Article 22 lacks clarity, we strongly disagree that it should be amended to limit its effect. We advocate amendment to strengthen its reach so that it applies to predominantly automated processing as well.

Individuals must be allowed to request information about algorithmic decisions which affect them and the data that is subject to decisions made.

It will be important that the public understand when and how AI intersects with their lives and what control they have over the process. Transparent and digestible explanations of AI systems and where they are employed will be helpful in demystifying the use of AI and promoting long-term trust in society.

The Liberal Democrats’ Science, Technology, and Innovation spokesperson Victoria Collins spoke in the second reading debate and you can read her speech here.

Overall, the Liberal Democrats support a modernised data framework that upholds digital rights while stimulating innovation. Our vision is of a digital future that spreads the benefits of technology across society while protecting fundamental liberties. However, we do need to see better safeguards in this Bill to protect fundamental rights and embed proper scrutiny. We will work with colleagues across the House, as well as with civil society organisations such as Big Brother Watch, Liberty and the Ada Lovelace Institute, to ensure that those important protections are not overlooked.

Thanks again for writing to me about this incredibly important issue.

Best wishes,

Layla

Layla Moran
Liberal Democrat Member of Parliament for Oxford West & Abingdon

Privacy concerns about the Data Use and Access Bill

2025-02-07T16:23:00.001+00:00

At the prompt of the Open Rights Group, I've written to Layla Moran about the government's proposed the Data Use and Access Bill.

Dear Layla Moran,

I am a resident of Abingdon getting in touch with you as the Liberal Democrat MP for Oxford West and Abingdon

I am writing to you with concerns about the Data Use and Access Bill. The Bill will have its second reading in the House of Commons on Wednesday, 12 February.

This Bill threatens to undo data protections that have safeguarded people's privacy and security for years. It is a significant move away from the EU's GDPR model towards a weaker, US-style system. This dangerous step could leave individuals vulnerable to data exploitation and harm. Businesses depend on a shared set of data protection standards under a common Human Rights framework with our closest trading partners. If we allow this to happen, we not only risk the UK's adequacy agreement with the EU—which is crucial for trade—but, more importantly, erode the fundamental rights of millions of people.

I want to live in a country where personal information is handled respectfully, not where corporations and the government can make decisions about us behind closed doors, using algorithms we can't possibly hope to understand. This Bill weakens key protections and removes essential safeguards, leaving people with little recourse if they are unfairly treated. Have we not learnt anything from recent scandals such as the Horizon Computer system scandal with postmasters? I urge you to stand against these dangerous provisions and fight for amendments that protect people's rights.

1. Safeguarding Rights Against Automated Decision-Making and AI

This Bill dismantles the right not to be subject to fully automated decisions (Clause 80), meaning AI systems could make significant choices about people's lives—including their job prospects, credit eligibility, or access to public services. We already know AI systems can reinforce discrimination and errors, so why weaken protections when we should strengthen them?

Recommendation: Drop Clause 80 and expand Article 22 protections to include increased safeguards around automated decision-making. This helps to ensure a safer roll-out of the expanding role of AI in Government.

2. Preventing Unaccountable Ministerial Powers Over Data Protection.

This Bill grants new secondary legislation powers to the Secretary of State (Clauses 70, 71, 74, 80, 85, and Schedule 7), allowing them to override protections with the limited scrutiny Statutory Instruments receive. This is an affront to democratic accountability. Open Rights Group warns in its latest report how this power could undermine integrity in UK elections, as any party in power could, at very short notice, change the rules on how they can use voters' data in elections. Also, recent events in the United States have shown how access to Government data can become a battlefield in the event of a constitutional crisis: giving unaccountable powers to the Secretary of State is a dangerous proposition that unnecessarily exposes us to the risk of seeing data we gave to a local authority or department being abused via Statutory Instruments.

Recommendation: Remove these Henry VIII powers and instead define what you want to do with data in primary statutory legislation.

3. Strengthening Accountability in Law Enforcement Data Sharing

The Bill lowers accountability for how police and other authorities access and use people's data. It removes the requirement for law enforcement to consider the impact of data-sharing on individuals (Schedules 4 and 5) and even eliminates the need for police to record why they are accessing a database (Clause 81). This makes abuse and overreach far more likely. The example of the Sarah Everard case, in which Police Officers were sacked for wrongful accessing case files, demonstrates why these safeguards are required.

Recommendation: Drop Schedules 4 and 5 and Clause 81 to restore accountability and transparency in law enforcement data use.

4. Addressing the ICO's Failure to Enforce Data Protection Laws.

The Information Commissioner's Office (ICO) is failing in its duty to enforce data protection laws. This is leaving people you represent in Oxford West & Abingdon —especially the most vulnerable—at risk of harm. Open Rights Group's Alternative Annual Report 2023-24 reveals that the ICO has repeatedly failed to act against serious breaches, particularly in cases affecting marginalised individuals. Instead of strengthening the ICO, this Bill weakens its independence, allowing more political influence over its work (Clauses 90 and 91, Schedule 14). This must not be allowed to happen.

Recommendation: Strengthen the ICO's independence by removing ministerial interference and increasing its enforcement powers. Also, the ICO's use of 'reprimands' should be limited, as these are ineffective methods of enforcing data protection laws.

Beyond individual rights, this Bill has significant economic implications. Losing the EU's adequacy agreement would be devastating to UK businesses, potentially costing between £1 billion and £1.6 billion in compliance costs and lost opportunities (New Economics Foundation and UCL European Institute). This is an unnecessary, self-inflicted wound that Parliament must prevent.

We need a data protection system that works for people—not just corporations. Please advocate for your constituents' rights by opposing these damaging provisions and working towards amendments that will restore public trust and accountability.
Yours sincerely,

Ray Corrigan

Note to MP on parliamentary debate on face recognition

2024-11-07T15:57:00.001+00:00

At the behest of Big Brother Watch, I've writen to my MP, Layla Moran, asking her to participate in the Westminster Hall debate on police use of face recognition technology scheduled in parliament for next Wednesday, 13 November, from 9:30am to 11am.

Dear Layla,
You may recall that I wrote to you last year noting a survey of over 100 MPs, carried out by YouGov on behalf of Privacy International, indicating that 70% of MPs did not know whether facial recognition technology (FRT) was being used in public spaces in their constituency.

On this occasion I am writing to you to ask you to attend and speak at a debate taking place next Wednesday 13th November at 9:30 am in Westminster Hall on the police’s use of live facial recognition. The debate is sponsored by Sir John Whittingdale MP and I would appreciate it if you could raise the concerns I have about the use of this technology and its impact on human rights and civil liberties in the UK.

Live facial recognition works by creating a 'faceprint' of everyone who passes in front of the camera — processing biometric data as sensitive as a fingerprint. This form of surveillance is deeply intrusive, often subjecting many thousands of innocent people to biometric identity checks without justification. Seven police forces around the UK are currently using this technology despite there being no specific law which governs its use, with forces such as Essex Police, the Metropolitan Police and South Wales Police continuing to use LFR on a regular basis in a way that affects millions of us as we live and travel around the UK.

Uses of live facial recognition have resulted in privacy and data breaches, misidentifications, racial discrimination, and a significant chilling effect on freedom of expression and assembly. In 2020, South Wales Police was found to have deployed live facial recognition surveillance unlawfully (Bridges v SWP) and the Metropolitan Police is also facing a judicial review brought by a Black victim of live facial recognition misidentification which took place earlier this year. Other victims of live facial recognition harassment and errors are initiating legal action in the retail context.

In recent years, parliamentarians across parties in Westminster, members of the Senedd, rights and equalities groups and technology experts across the globe have called for a stop to the use of this technology. The only detailed inquiry into the use of live facial recognition by a parliamentary committee called for a stop to its use. The Equality and Human Rights Commission (EHRC) called for the UK Government to suspend live facial recognition in 2020.

Governments across the democratic world are legislating to ban and significantly restrict the use of live facial recognition surveillance, for both law enforcement and private companies – but successive governments have left the UK behind. The EU’s AI Act will introduce an almost total ban on the use of live facial recognition, with law enforcement exceptions for only the most serious crimes and on the condition of prior judicial approval for each deployment. In the US, multiple states have banned law enforcement from using the technology entirely. The UK risks becoming an outlier in the democratic world, instead following the approach of countries like Russia and China, which have heavily invested in this technology to the detriment of their citizens’ rights and freedoms.

The legal vacuum when it comes to the police’s use of live facial recognition technology cannot continue. The UK must follow the example set by European states and introduce stringent restrictions on the use of this surveillance technology akin to the EU’s AI Act. For more information, please contact Big Brother Watch at info@bigbrotherwatch.org.uk
In addition to police use of such technologies, there is an increasing expansion of their deployment in the retail sector and other areas. The local Budgens supermarket has recently installed a face recognition system. Having used that shop for 25 years, I intend never to cross their threshold again. For ordinary citizens to have to exclude themselves from routine activities and locations they frequented for generations, simply to avoid these intrusive surveillance technologies is not conducive to the maintenance of a healthy society.

As my MP I would appreciate it if you could attend next Wednesday’s debate and raise my concerns.

I look forward to hearing from you.

Kind regards,
Ray Corrigan

Proposed reforms to the National Planning Policy Framework and other changes to the planning system

2024-09-20T16:28:00.001+01:00

I've responded to the government consultation on Proposed reforms to the National Planning Policy Framework and other changes to the planning system via Foxglove.

Dear Planning Policy Consultation Team,

I am an individual responding to the government’s consultation on ‘Proposed reforms to the National Planning Policy Framework and other changes to the planning system’ (question 64).

Thank you for giving the British public the opportunity to have their say on these plans.

I want to register my concerns as an individual about the potentially devastating impact of new data centres on the environment, on water and electricity production and use, and on the UK’s transition to Net Zero and green energy.

The companies building these data centres are some of the richest on the planet. The government must make their involvement conditional on their investment in green energy – ensuring they give more than they take for the UK and from the environment.

The impacts can and should be mitigated by the planning application process in the following straightforward and commonsense ways:

1. Companies building new data centres must be able to show that the centre will have no negative impact on our targets to avoid climate catastrophe, including Net Zero by 2023.

2. Companies building new data centres must be able to show that the centre will not place significant additional burdens on our power and water supply.

3. Companies building new data centres should provide a plan explaining how it will sustainably meet the power and water needs of the centre without burdening the UK’s existing power and water supplies. Any costs associated with the additional energy or water requirements of a new data centre application must be covered by them - not by the British taxpayers.

Thank you,

Ray Corrigan

Twitter has locked me out...

2024-07-19T16:38:00.000+01:00

... and I'm not sure I want to even bother giving Mr Musk a fictional age to be allowed back in.

Reform response to ORG call to respect digital rights

2024-06-28T07:58:00.000+01:00

The first response to my email to parliamentary candidates asking them to support digital rights in the next parliament has come from Reform candidate, James Gunn. He says:

"Hi Ray,

Thanks for your email, This all makes a lot of sense although I might allow advertisees to optin for tracking for free services subject to a proper explanation of what was being tracked and to whom it would be sold.

Here's "Our Contract with You" - like a manifesto, but we consider it a binding contract! This is my only pledge

https://assets.nationbuilder.com/reformuk/pages/253/attachments/original/1718625371/Reform_UK_Our_Contract_with_You.pdf?1718625371

I hope you’ll vote for me / Reform on 4th July

BWs
James

James Gunn
Parliamentary Candidate Reform, Oxford West and Abingdon
https://gofund.me/e14a0c69"

Calling on election candidates to support digital rights

2024-06-27T16:39:00.002+01:00

I have written to my prospective parliamentary candidates:

Layla Moran - Liberal Democrats

Stephen Webb - Labour

Chris Goodhall - Green Party

James Gunn - Reform

Anni Byard - Social Democratic Party

Ian Oliver Michael Shelley - Christian People's Alliance

... asking that they commit to support a core selection of digital rights that the Open Rights Group are campaigning for.

"I'm writing to you as a [ ] candidate in the general election, as I care about digital rights.

I am a supporter of Open Rights Group a UK-based digital rights organisation that campaigns to protect our rights to privacy and free speech online.

"Open Rights Group is calling on the next government to:

1. Protect our right to send secure messages
Everyone – including children and young people – should have the right to use end-to-end encryption to ensure that our communications are safe, secure and private. The next government should protect not undermine encryption.

2. Provide migrants with digital sanctuary
Migrants, refugees and asylum seekers need the same data and privacy rights as everyone else so that they can keep their digital identity and information safe. The next government should commit to ending the digital hostile environment.

3. Ban the use of pre-crime AI by the police
Predictive policing systems that use artificial intelligence (AI) to ‘predict’ criminal behaviour undermine our right to be presumed innocent and exacerbate discrimination and inequality in our criminal justice system. The next government should ban dangerous uses of AI in policing.

4. Defend our right to freedom of expression online
Freedom of expression online is being undermined by age verification, content takedowns, social media censorship and unfair copyright claims. The next government should commit to protecting our right to freedom of expression online.

5. Strengthen our data protection rights
We need strong data protection laws to make sure that governments and companies do not use our data to track, surveil and profit from us. The next government should strengthen our data protection rights and make sure that the data protection watchdog is fit for purpose.

6. End intrusive tracking by online advertisers
Advertising companies track our internet use to build detailed profiles so they can target us with adverts. The next government should restrict intrusive tracking by data brokers and online advertisers.

We know that Political Parties will often not allow candidates to agree to support particular policy positions outside their manifesto.

Therefore, I am instead asking you to make a simple pledge. That, if elected, you will uphold digital rights in the next parliament.

If you want to register your support as a candidate that is committed to digital rights, then ORG has produced a page where you can do so - https://action.openrightsgroup.org/i-pledge-support-digital-rights

If you want to read ORG's full Digital Rights 24 Manifesto this can be read at. - https://www.openrightsgroup.org/publications/digital-rights-24/

I look forward to hearing your views on digital rights."

Reply from MP on Investigatory Powers (Amendment) Bill

2024-05-01T17:01:00.000+01:00

I got a response from my MP, Layla Moran, relating to concerns about the Investigatory Powers Amendment Bill, now Investigatory Powers (Amendment) Act 2024. It's boilerplate generic which, given Ms Moran's personal concerns on the terrible circumstances in Gaza, is understandable.

"Dear Ray

Thank you for writing to me about the Investigatory Powers Bill.

In a changing world we rely on our intelligence services to help keep us safe. And we need a police force (National Crime Agency) that can cope with the internationalisation of crime. Liberal Democrats support the services that seek to do this.

These vital tasks have to be balanced against the freedoms and liberties at the heart of our country’s values. Every new power must be weighed in that balance.

Liberal Democrats are concerned about how these changes could undermine the privacy of everyone.

We have particular concerns about reductions in judicial scrutiny, and the possible use of available medical, genetic and legally privileged data.

For all its powers our police and security services need resources to take on the multinational cartels that prey on our people and threaten national security. Liberal Democrats will continue to push for resources to target these threats to our way of life.

Thanks again for writing to me about this incredibly important issue.

Best wishes,

Layla

Layla Moran
Liberal Democrat Member of Parliament for Oxford West & Abingdon"

On the Open Rights Group Report - PREVENT and the Pre-Crime State: How unaccountable data sharing is harming a generation

2024-03-07T14:00:00.002+00:00

At the request of the Open Rights Group I have written to my local councillor, drawing their attention to the excellent report ORG have published on the Prevent programme, Prevent and the Pre-Crime State: How unaccountable data sharing is harming a generation.

"I am writing to bring to your attention to a recent report by the Open Rights Group, revealing concerning issues related to the UK's counter-terrorism Prevent programme. Prevent operates in the pre-crime space, in which no offence has taken place, but rather people are surveilled and viewed as suspicious. It operates by extracting data and policing information that further securitises the spaces of marginalised and vulnerable communities.

I believe it is crucial for us to ensure that the implementation of such programmes by our local authority aligns with the principles of transparency, accountability, and respect for individual rights.

The report highlights widespread data sharing and retention of Prevent referrals, even when marked 'no further action.'

Thousands of Prevent referrals are made each year, and most of these referrals concern children. The majority do not meet the threshold for a Channel intervention. Despite this, the data of Prevent referees can be retained for at least six and up to 100 years, raising potential privacy and civil liberties concerns.

Key findings from the report include:

- Referrals are stored within a national Prevent database, irrespective of whether they meet the threshold for a Channel panel review.

- Data is retained for a minimum of six years, with the possibility of being kept for up to 100 years, even when a case is marked no further action. This potentially violates individuals' rights if there is no policing purpose for retention.

- It is very difficult for individuals to exercise their right to erasure and request data is removed because many will not know that they have been referred to Prevent. Even when they do know, the lack of transparency about data sharing makes it very difficult for individuals to find out all the different places that their data is being held.

-ORG is calling for local authorities, police departments and individual institutions subject to the Prevent duty to ensure maximum transparency around referrals, data processing and data sharing practices, including the systems used and in as clear detail as possible.

- Data of Prevent referees is being shared with airports, ports, and immigration services, leading to real-life harms on people' lives.

- Particularly concerning is the impact on children, who make up the majority of Prevent referrals, whose data is being shared with children's services and retained for an extended period.

Given these findings, I am reaching out to request that this report and its recommendations are considered by the local authority. Could you please request it is placed on the agenda of an appropriate meeting.

It is essential to ensure that our community's rights and privacy are respected, and that the program's implementation strikes the right balance between security measures and protecting individual freedoms.

The report also includes case studies of children experiencing harms as a result of being referred to Prevent, underlining the urgency of addressing these issues.

As a Councillor with a commitment to our community's well-being, your involvement in reviewing and potentially advocating for improvements in the local handling of Prevent would be highly valued.

I understand your time is precious, but I believe that your attention to this matter can contribute to fostering a safe and just community. If you are open to it the report can be downloaded and read at the following website address - https://www.openrightsgroup.org/publications/prevent-and-the-pre-crime-state-how-unaccountable-data-sharing-is-harming-a-generation/

Thank you for your dedication to serving our community, and I look forward to hearing your thoughts on this important issue.

Yours sincerely,
Ray Corrigan"

Note to MP on DWP algorithm targeting disabled people for fraud investigation

2024-03-05T13:11:00.001+00:00

At the behest of Foxglove, I have written to my MP, Layla Moran, about a discriminatory DWP algorithm being used to falsely accuse disabled people of benefit fraud.

"Dear Layla,

I’m very concerned about the Department for Work and Pensions (DWP) use of an algorithm that targets disabled people for benefit fraud investigations.

Disabled people describe being flagged by the algorithm for no obvious reason and cast under suspicion of benefit fraud. They are then subjected to a lengthy, aggressive, and invasive investigation process.

I understand that UK law requires the DWP to test for and ensure safeguards are in place to address bias and discrimination in all its automated technologies. Despite that, the DWP has chosen not to test this algorithm(s) for potential impact on individuals with disabilities.

This is very concerning and there have been too many reports of innocent disabled people subjected to unfair, humiliating, distressing and invasive investigations because of this algorithm. This is particularly concerning after the Post Office Horizon scandal.

Please can you seek answers from the DWP? I would like you, as my MP, to write to the minister in charge and ask them:

· to explain in full how the system works
· what the DWP are doing to ensure it's not discriminatory.

I think if they can’t prove it’s not harming innocent people, then they should stop using it.
Yours sincerely
Ray Corrigan"

Reply from MP on face recognition technologies

2024-01-08T18:00:00.000+00:00

Layla Moran MP has responded to my concerns about the expanding deployment face recognition technologies :

22 December, 2023

Dear Ray,

Thank you for writing to me, and for your patience as I needed to focus on urgent matters in Gaza.

I have written to the Chief Inspector to raise your concerns, as you can read below. When I receive an update I will pass that on to you.

In general, I am deeply concerned by any laws that unnecessarily erode civil liberties which is why we want to:

End the bulk collection of communications data and internet connection records.
Immediately halt the use of facial recognition surveillance by the police.
Introduce a Code of Ethics to govern the use of personal data and artificial intelligence to ensure it is unbiased, transparent and accurate, and respects privacy.

Thank you again for raising this concern with me, let’s see what the Chief Constable says.

Best wishes,

Layla

Layla Moran
Liberal Democrat Member of Parliament for Oxford West & Abingdon

________________________________________

Dear Chief Constable,

Facial Recognition Technology

My constituent, Ray Corrigan, has expressed his concern about the use of Facial Recognition Technology (FRT) being used locally by Thames Valley Police.

In particular, Ray asks how widespread the deployment of FRT is in the Abingdon and Oxford areas. Ray tells me he is concerned about the risk to privacy, particularly of private citizens being tracked as they go about their lives.

Could you please outline your policy, and how it accounts for this concern. Could you also outline the intended benefits of any FRT being used, and what alternatives were considered.

Thank you for your attention on this matter.

Best wishes,

Layla

Layla Moran
Liberal Democrat Member of Parliament for Oxford West & Abingdo

She had a response from Thames Valley Police:

5 January, 2024

Dear Ray,

Happy New Year, I hope you had a restful holidays?

I have received a response from Thames Valley Police regarding Facial Recognition Technology being used in our area. I hope this reassures you that powers are not being abused and that there are no unacceptable intrusions on our privacy via FRT

Best wishes,

Layla

Layla Moran
Liberal Democrat Member of Parliament for Oxford West & Abingdon

________________________________________
From: Chief Constable TVP
Sent: 27 December 2023 15:11
To: "MORAN, Layla" <layla.moran.mp@parliament.uk>
Subject: RE: Facial Recognition Technology (Case Ref: LM62195)

Dear Layla

I hope you have managed to enjoy a good Christmas break so far.

In regards to the below enquiry on behalf of your constituent Mr Ray Corrigan, I hope I am able to assist by responding on behalf of the Chief Constable, to avoid an unnecessary delay.

In terms of Facial Recognition we are only making use of the Police National Database facial search facility (Retrospective Facial Recognition), something available to other Forces.

We have no other facial recognition facility in use across any of the Force area.

The below link provides an explanation of the national solutions available but, other than the Police National Database facial search option detailed in the first section, currently we have not adopted any other elements of this capability:

Police use of Facial Recognition: Factsheet - Home Office in the media (blog.gov.uk)

I hope this enables you to provide the necessary response to your constituent.

Wishing you a safe and prosperous 2024.

Yours sincerely

[Embedded Image] [Embedded Image]

On behalf of Jason Hogg, Chief Constable

(:

7:

+: Thames Valley Police HQ

Oxford Road, Kidlington, OX5 2NX

protecting our communities