SpaceX had launched its latest batch of Starlink satellites on a Falcon 9 rocket from Cape Canaveral in Florida on Thursday, February 3. This was SpaceX’s 38th Starlink launch; in all, the company has launched more than 1,900 of the car-size satellites, and eventually it wants to have up to 42,000 of them in low Earth orbit to deliver the internet to all corners of the globe.

The day after the launch, however, disaster struck. An eruption of plasma from the sun sent charged particles streaming into Earth’s atmosphere, sending the planet’s magnetic field haywire and increasing the density of its atmosphere. That increase in density meant there were more particles to push against satellites in Earth’s orbit. This phenomenon, known as atmospheric drag, can pull them out of their orbital paths.  

To support MIT Technology Review’s journalism, please consider becoming a subscriber.

As a result of the storm, as many as 40 of the new satellites “will reenter or already have reentered the Earth’s atmosphere,” SpaceX said in a statement, describing it as a “unique situation.” These satellites were vulnerable because they are launched into a low orbit, between 210 and 240 kilometers, where the atmosphere is denser, making the effects of the storm worse. The satellites are meant to use onboard ion thrusters to slowly raise their orbits to 550 kilometers over several weeks. Those already in these higher orbits were less affected because the atmosphere is much thinner at that altitude, so drag is reduced.

SpaceX noted that the satellites were designed to completely burn up in the atmosphere, “meaning no orbital debris is created and no satellite parts hit the ground.” A handful of the satellites have already reentered, and the rest are expected to do so within a week. But the financial cost of the botched launch is estimated to be between $50 millionand $100 million.

And the event has raised some important questions about the planned rollout and future of mega-constellations. The National Oceanic and Atmospheric Administration (NOAA) had warned of the possibility of a geomagnetic storm days before the launch, yet SpaceX decided to go ahead anyway. Experts are not sure why. “It is a bit weird,” says Marco Langbroek, an astronomer at Leiden University. “Maybe they did not expect the effects to be this large.” 

In fact, the storm ranked as a relatively minor G1 on a scale that runs from G1 to G5. While SpaceX said this caused atmospheric drag “to increase up to 50% higher than during previous launches,” the effect was still relatively small. More extreme events can be much more dramatic. “This storm itself was not particularly large,” says Delores Knipp, a space weather expert at the University of Colorado, Boulder. “We’ve seen the atmosphere expand 1,000%. You can get a 10 times increase in density at various altitudes.”

Those bigger effects could come into play relatively soon because the sun is expected to reach the peak of its 11-year activity cycle, known as solar maximum, in 2025. This will make powerful eruptions and geomagnetic storms more common. “There are reasons to be concerned,” says Knipp. “These expansions of the atmosphere will happen on an irregular basis as we move into solar maximum.”

That the Starlink satellites were unable to overcome even a minor storm suggests SpaceX needs to approach future launches differently. It may need to deploy the satellites at a higher altitude, where the atmosphere is thinner, to ensure that they won’t be pushed out of orbit. “Three hundred kilometers should be enough,” says Jonathan McDowell, an astrophysicist from the Harvard-Smithsonian Center for Astrophysics. That could result in “at most a 10% increase in launch costs,” says McDowell.

In turn, that could slightly affect the rollout speed of Starlink: the company would need to fly fewer satellites per launch so that each would have enough fuel to reach higher altitudes. It also means any satellites that malfunction will take longer to reenter Earth’s atmosphere, diminishing what SpaceX had touted as a benefit of launching to lower altitudes: this was supposed to minimize space debris because failed satellites would fall back to Earth more quickly. “It’s a trade-off,” says Hugh Lewis, a satellite expert from the University of Southampton. At 200 kilometers, a dead satellite will stay in orbit for “days at most,” says Lewis, but that period rises to several weeks at 300 kilometers and above.

Managing these mega-constellations could be a problem too. While we have experienced solar maximum with satellites in orbit before, the number orbiting now is unprecedented. By 2025, there could be more than 10,000, not only from SpaceX but from other ventures such as Amazon’s Project Kuiper and the UK’s OneWeb. Future storms could frequently push and pull these satellites, changing their positions and putting them at risk of colliding.

“We’re talking about kilometers in terms of altitude being changed,” says Lewis. “The more satellites that go into orbit, our ability to manage that complexity is going to be limited. At some point, we’re going to see something more severe happening than just 40 satellites reentering.”

Amazon said its constellation, and the design of the satellites themselves, had been designed to cope with this increased solar activity but did not provide specific details. SpaceX and OneWeb did not respond to a request for comment.

This latest event highlights how carefully all mega-constellation operators will need to plan for the effects of solar activity, since any collisions could add thousands more pieces of space debris that could affect our ability to use Earth’s orbit safely. “I have to believe they’ve factored it into their plans,” says McDowell. “Maybe they missed this particular issue, but they have to have run their models, one hopes.”

What is certain is that we are heading into uncharted waters. “This region [of orbit] we’re talking about is so valuable and important,” says Lewis. “Everybody needs to do a much better job of using foresight to anticipate these issues.”

The crowd was gathered because a team of international disease detectives selected by the World Health Organization (WHO) to hunt for the origins of covid-19 was on its way to visit. 

“They will be here in a minute,” a journalist working for Japan’s Tokyo Broadcasting System Television said after checking her phone. Her voice was brisk and slightly shaken; her eyes sparkled with nervous excitement. “My colleagues just told me. They’re chasing the WHO cars.” 

Soon enough, the motorcade burst through the fog. As it approached the institute’s main gate, a journalist in a blue down jacket and white face mask sprinted alongside as if he were running for his life, pointing a video camera toward the cars, his rucksack bouncing up and down on his back. A dozen photographers flocked to the lead car, pushing against one another and forcing the convoy to a stop. The guards tried herding them away to get the cars moving again. “Comments, please!” several journalists shouted. 

Inside the car, Peter Daszak—a disease ecologist and president of the EcoHealth Alliance, a New York-based nonprofit that works with scientists around the world to study viruses in wildlife—was filming the scene on his cell phone. 

NHUNG LE

He was a member of the WHO team, and when we’d spoken the week before, he’d cautioned that the Wuhan trip was just a first step in trying to figure out where covid-19 came from. “It can take years or even decades to find the cause of a new infectious disease,” said Daszak, who has collaborated with the Wuhan Institute of Virology for more than 15 years and is now himself caught up in the debate over the disease’s origins. “Sometimes we just never know.” 

But the world wanted quick answers.

The institute holds a critical place in the story of the covid-19 pandemic. A leading center for coronavirus research, it was the first facility to isolate the new virus, and the first to sequence its genome. One of its labs, led by virologist Shi Zhengli, focuses on coronaviruses that live in bats, and has spent years sequencing viral genomes, isolating live viruses, and—through genetic mixing and matching—trying to understand how they may evolve to gain the ability to infect humans. Over the past 18 years, her team has collected more than 20,000 samples from bat colonies across China.

Shi’s work, which has earned her the nickname China’s bat woman, has been at the center of controversy. Some have suggested that her bat samples could be the source of the covid-19 virus, which scientists call SARS-CoV-2. They have claimed that the virus could have hitched a ride to Wuhan by infecting one of her team members in their fieldwork collecting samples from bats. Or, some speculate, the live viruses her team cultured in the lab, including—more worryingly—the ones they created by genetic tinkering, could be the source of the pandemic.

All eyes were on the WHO, the leading international public health agency, to probe covid-19’s origins. The team’s mission was to examine when and where the outbreak had started and how the new virus crossed over to humans. The report, which was released last March, concluded it was “extremely unlikely” that covid-19 could have been caused by a lab accident. The situation the team ranked most likely was that it had jumped from bats to humans via some intermediary animal. Their results, supported by research published in peer-reviewed journals and by ongoing studies, suggest that the pandemic probably started at the Huanan Seafood Wholesale Market in central Wuhan, where live mammals were sold and where most of the early covid-19 cases emerged.

Not everybody agrees, but the majority of virologists and infectious-disease experts, especially those working directly on the origins question, lean toward that theory, barring the emergence of new evidence that persuades them otherwise.

Spillover from animals to humans “was how almost every major epidemic got started in the past decades,” says Shi’s longtime collaborator Linfa Wang, an expert on emerging infectious diseases at the Duke–National University of Singapore Medical School and a member of the WHO team that in 2003 investigated the origins of SARS, a deadly infectious disease caused by a coronavirus now known as SARS-CoV-1. That illness sickened 8,000 people worldwide and killed nearly 800 between 2002 and 2004. “It’s a common and well-documented pathway,” he says.

But one year after the WHO’s visit to Wuhan, the disease detectives have yet to find the guilty animal or other indisputable evidence of natural origins. Critics also question the conclusion of the agency’s mission team partly because one of its members, Daszak, who is a prominent advocate of the natural origins theory, has potential conflicts of interest. Speculation over the possibility of a lab accident has surged. Inflaming the suspicions are concerns over biosafety procedures at the Wuhan lab, political tensions between China and the US, and a general sense that the Chinese government is not to be trusted. 

By trying to understand the process and context of Shi’s work—and to find out who she was—I wanted to learn what role, if any, China’s bat woman had in the origins of covid-19.

Scientists like David Relman, an expert on microbiology and biosecurity at Stanford University, are dismayed at the way the lab leak theory has been dismissed. He helped organize a group of 18 scientists to sign a letter published in Science last May calling for further investigation of a possible accident. (At least two of those involved later sought to distance themselves from the letter after seeing how it had been used to promote the lab leak theory.) Soon afterwards, President Joe Biden directed the US intelligence community to intensify its probe into the pandemic’s origins. The declassified report released in October shows that it reached no firm conclusion.

In December 2020, a month before the WHO visit, I too embarked on a search for answers. I talked to dozens of top scientists and biosafety experts worldwide. I spent six weeks in Wuhan, where I interviewed Shi and her team for a total of more than 40 hours. I had a private meeting with three members of the WHO mission. I visited the Wuhan Institute of Virology half a dozen times, often on the spur of the moment, and went with the scientists on a virus-sampling trip to a bat cave. By trying to understand the process and context of Shi’s work—and to find out who she was—I wanted to learn what role, if any, China’s bat woman had in the origins of covid-19.

Probing covid-19’s origins will not only help us understand how coronaviruses work but shine a bright light on the human behaviors—including the types of scientific research—that risk causing a pandemic in the future.

Like the WHO team, I have not gone through Shi’s freezers or lab books, and therefore I cannot prove or disprove whether activities associated with her research caused the pandemic. It’s more about providing additional perspectives—having Shi and her team tell their side of the story on the record, and in the most detail to date, so that the world can better understand how this deeply entrenched controversy has come about and how we can move forward.

Meeting China’s bat woman

I met Shi Zhengli in person for the first time on a cold afternoon in December 2020. We had spoken earlier that year for an article published in Scientific American. The level of access she has given me is unparalleled. She rarely speaks with the press, and her interaction with journalists writing for the Western media has been largely confined to emails and texts. She told me she spoke to me because my strong science background allows me to grasp the nuances and complexity of her work, because I understand China, and because we can communicate in Chinese, our native tongue, in which I conducted the interviews.

We met for lunch and then went for a walk in a nearby park. A few days later, I visited the institute’s city campus in central Wuhan—approximately 12 miles from the suburban site that the WHO team later toured. Her lab was on the second floor of a solemn-looking cream-colored building. The main room had rows of benches with weighing machines, polystyrene ice boxes, and desktop centrifuges. Bottles of chemicals and solutions were tightly packed on the shelves. One student was typing away on a computer, while another was pipetting a tiny amount of colorless liquid from one test tube to another. The scene gave me a sense of déjà vu—I’d spent a decade working as a molecular biologist, including six years as a postdoc. It reminded me of my days in the lab. 

“It’s probably not that different from where you worked,” said Shi, as if she could read my mind.

Shi is petite, with short wavy hair that is neatly combed. Her voice is high and light, with the sparkle of a soprano (she is an amateur folksinger). That day she wore a beige sweater and blue jeans. As we went on to other parts of her lab—the deep freezers that held bat samples, and the rooms for culturing cells in petri dishes—she explained that her team had about three dozen researchers. That’s a lot for a Chinese lab, but it’s not the gigantic operation that many outsiders imagine. “I do not have an army of researchers and unlimited resources,” she said. Until the pandemic hit, coronavirus research was not a trendy subject and could not easily attract funding. 

Shi is one of the rare breed of virologists who are just as comfortable in the field as in the lab. She grew up in a small village in central China’s Henan province and spent most of her childhood roaming the hills. She doesn’t regard herself as ambitious. When she graduated from the prestigious Wuhan University in late 1987, she told me, “I thought I had achieved my career goal and the next stage was to get married and have kids.” The main reason she went on to study at the Wuhan Institute of Virology was to stay in the same city as her then boyfriend. But as China invested in sending promising young scientists abroad to pursue doctoral degrees, Shi grabbed the opportunity.

In 2000, she got her PhD at Université Montpellier 2 in France. Studying there was an unusual decision since she didn’t speak French, and a difficult one because it meant leaving her young son behind in China; the stipend was not enough to support a young family. But the experience left a positive imprint; she particularly appreciated the Western culture that prized “critical thinking, independent-mindedness, and not following the crowd,” she told me. “You can’t do great science without any of these. This is what China really needs to get better at.”

Afterwards, she returned to the Wuhan institute, where she focused mainly on aquaculture pests until 2004. At that time, the world was still reeling from SARS, and Wang, the Duke-NUS infectious-disease specialist, was working in Australia and looking for a virologist in China to help hunt for the origins of the new disease. Shi jumped at the opportunity, joining an international team to collect blood, urine, saliva, and feces from bat colonies in mountainous areas across China. They found SARS-like coronaviruses in bats within a year, but it took nearly a decade to prove that bats were the source of the contagion. Through their collaboration, Shi and Wang became friends; colleagues knew them for their karaoke duets, and they earned the nicknames “bat woman” and “bat man,” respectively. 

To support MIT Technology Review’s journalism, please consider becoming a subscriber.

As Shi showed me around her lab, she pointed to the deep freezers where the team kept tens of thousands of bat samples in chemical soups. She told me how virus-containing samples are kept frozen in the field, either on dry ice or in liquid nitrogen, before being transferred to dedicated, double-locked deep freezers in the Wuhan lab. Only designated personnel can access those samples; they need approval from two senior staff members, each of whom is in charge of a separate key to the two locks. All access to the samples is logged.

The core of her research over the past 18 years, she explained, has been to look for bat viruses that are closely related to SARS-CoV-1, and to understand how they could evolve new features that allow them to infect humans. She talked me through that process, which begins with testing each bat sample to see if it contains a coronavirus—using the same PCR-based technique as many covid-19 tests. All coronaviruses contain a gene that encodes an enzyme called RNA-dependent RNA polymerase, or RdRp, which helps viruses replicate by making more copies of their genomes. If the characteristic RdRp shows up in a bat sample, it’s a telltale sign that a coronavirus is present. 

At first glance I was concerned by the sheer size of Shi’s collection of more than 20,000 bat samples. But she explained that on average only 10% contain coronaviruses, and only 10% of those are closely related to SARS-CoV-1: in all its years, the team has identified approximately 220 such viruses. The findings, say virologists such as Edward Holmes of the University of Sydney, have provided valuable insight into the evolutionary history of coronaviruses and the way they generate genetic variants.

Whenever the team found a bat relative of SARS-CoV-1, Shi says, she asked the same questions: How threatening is it to other animal species, including humans? What would it take for the virus to become one that, like SARS-CoV-1, can cause major epidemics?

The real thing

An important way to test if a coronavirus can evolve into something more threatening is to see whether its spike proteins—the weapons of invasion that give the virus a crown-like appearance—can latch onto a molecule called angiotensin-converting enzyme 2, or ACE2, which is present on the surface of cells in most vertebrates. To find out about a virus’s potential to infect people, Shi’s team would sequence its spike gene, compare it with that of SARS-CoV-1, and study on a computer its structure and ability to bind to ACE2. 

The researchers also used pseudoviruses—viruses whose ability to copy their genomes is disabled—to test whether the spikes could help them enter cells from various animals. Scientists all over the world use this approach to study new pathogens without resorting to live viruses. It can be conducted with relatively inexpensive biocontainment precautions at what’s known as biosafety level 2, or BSL-2: researchers wear gloves and lab coats, and they work in cabinets that have air filtration and are under negative pressure to keep pathogens inside.

The first step for this type of work is to extract genetic material for genomic sequencing, which would inactivate all the microbes in the sample. This and subsequent cell-entry studies using pseudoviruses are well-established, safe methods.

But while pseudoviruses are a great tool, spikes—it’s become increasingly clear—are not the only factor that determines a virus’s ability to infect cells. The approach also can’t show, for instance, how exactly a virus makes cells sick, how it spreads from one cell to another, or how a pathogen might evade the body’s immune response. Those questions, which are critical for the development of drugs and vaccines, can be addressed only by using the real thing—a fully functional virus. And it’s this more dangerous work that has become the center of the controversy around Shi.

Isolating live coronaviruses from bat samples is notoriously tricky—mostly because only a small fraction of samples contain even a whiff of the viruses (whereas specimens from people with SARS or covid-19 are often teeming with coronaviruses). The process of culturing viruses involves providing them with cells they can infect. Several labs around the world have tried to get live bat coronaviruses and failed. Until January 2021, the Wuhan lab was the only one that had managed the feat, according to Stephen Goldstein, a coronavirus expert at the University of Utah in Salt Lake City. And the person with the green thumb was Yang Xinglou, a senior research scientist on Shi’s team.

I met Yang at the institute’s maximum-biocontainment campus on the outskirts of Wuhan on a muggy afternoon last May. He came to pick me up at the main gate wearing a turquoise-colored T-shirt and jeans. In his mid-30s, Yang was slim and of medium height. His hair was neatly trimmed, but in a sudden breeze, his bangs danced over a forehead dominated by thick brows. I filled out a registration form and showed the security guards my national identity card, and we walked to his office across the neatly manicured campus. 

Instead of walking along the winding, camera-lined driveway meant for cars, we stepped onto a narrow path that ran by a small lake. On the far side I could see an austere-looking square building, about four floors high, sturdy, with silver siding and few windows. Inside it was China’s flagship BSL-4 lab—the crown jewel of the country’s microbiology work.

I did not go inside the BSL-4 facility: there are strict protocols that make it difficult for any visitors to get in, let alone the press. I did, however, visit the nearby BSL-3 lab, which handles less deadly pathogens. After undergoing further security checks, we entered its control room, where large screens revealed what was inside: a preparatory room, three rooms for culturing cells, a room for working with small animals such as mice and rats, a dedicated space for disinfection, and the entrances to both the lab and the control room itself. While I watched, one researcher put materials into a decontamination chamber, and two scientists in white full-body protective suits sat in front of a biosafety cabinet, working with rows of small vials behind a glass screen. A black tube on the back of their suits delivered filtered air to their face masks. 

It was here, on January 5, 2020, that Yang first successfully isolated SARS-CoV-2 from a patient sample—the first isolate of the new coronavirus. “Which room did you use?” I asked. “Cell culture room 3,” he told me, pointing at one of the screens. “It was in that cabinet.” 

It was just an ordinary cabinet in an ordinary room, with two bottles of disinfectant and two biohazard garbage bins behind the glass screen—but it’s now a landmark in the battle against the biggest pandemic in a century. 

It was here, on January 5, 2020, that Yang first successfully isolated SARS-CoV-2 from a patient sample—the first isolate of the new coronavirus.

Yang has worked at the institute with pathogens in bats and rodents since 2008, developing and refining virus-catching techniques. There were lots of failures along the way, but in 2012, he hit the jackpot: a sample his team retrieved from a bat cave near Kunming successfully infected a type of monkey kidney cells called Vero E6, which has high levels of ACE2 on its surface. Once a live virus was at their disposal, the scientists could test directly whether it posed a potential threat.

It was a major breakthrough: for the first time researchers were able to demonstrate that bat coronaviruses in a petri dish could also infect cells from other species, including pigs and humans, by binding to their ACE2 receptors. The virus was 95% identical to SARS-CoV-1. The team named it WIV1 to indicate that it was isolated at the Wuhan Institute of Virology. Their study, published in Nature in 2013, provided strong evidence that SARS-CoV-1 originated in bats. 

In all his years of work, Yang has managed to isolate only three bat coronaviruses—all of them close relatives of SARS-CoV-1. More recently, the team managed to synthesize three bat coronaviruses from their genomic sequences. All six are close relatives of SARS-CoV-1. None of them, said virologists MIT Technology Review spoke to, could have been the source of SARS-CoV-2: they’re just too different. 

There was, however, one other virus in a bat sample that is a lot closer to SARS-CoV-2—96% identical. It has its own curious origin story, and in some parts of the scientific community and beyond, it’s become a prime suspect in the hunt for the pandemic’s origins. It’s called RaTG13.

Mine mystery

In late April 2012, a strange disease emerged from an abandoned copper mine near the town of Tongguan in Mojiang county, a region in China’s southwestern province of Yunnan. Six workers who had been cleaning up bat guano in the mine fell ill with pneumonia-like symptoms—coughs, headaches, fevers, and aching limbs—and were admitted to a hospital in Kunming, the provincial capital. One died in 12 days, and two recovered in a month, followed by another death on June 12. 

A week later, the country’s leading respiratory clinician, Zhong Nanshan, joined a clinical consultation remotely with colleagues at the Kunming hospital to determine how to treat the remaining two Mojiang patients. Zhong, the former director of the Guangzhou Institute of Respiratory Diseases, had played an instrumental role in the fight against SARS. He noted that the miners’ lab tests and CT scans were uncannily similar to those of patients with SARS, which hadn’t been seen since 2004. The clinicians in Kunming, he told me, suspected that a fungus had caused their illness—because cave-associated fungal infections happen in Yunnan every now and then—but Zhong thought a virus might be involved. He asked Shi’s team to test the patient samples for viral infections, but they couldn’t find any evidence of infection by coronaviruses or other known viruses. 

In 2020, with the pandemic raging, some scientists—including Stanford’s Relman—wondered if Shi had been wrong. Perhaps, they say, a SARS-like coronavirus was to blame. Perhaps there was even a link between the disease that affected the Mojiang miners and covid-19. 

That suspicion was bolstered in May 2020, when the anonymous owner of the Twitter handle @TheSeeker268—who claimed to me in Twitter texts that he is a 30-year-old man trained in architecture and filmmaking and lives in the Indian city of Bhubaneswar—dug up a 2016 PhD thesis by Huang Canping from the Chinese internet. Huang was a student of George Gao, director-general of the Chinese Center for Disease Control and Prevention in Beijing, and his thesis cited the Wuhan Institute of Virology as claiming that four of the Mojiang miners had antibodies against SARS-CoV-1. Scientists like Monali Rahalkar, a microbial ecologist at the MACS Agharkar Research Institute in Pune, India, and a strong proponent of the lab leak theory, said that this suggests the miners were infected by a SARS-like coronavirus. Social media and the press teemed with suspicion that Shi tried to hide the fact.

SCIENCE PHOTO LIBRARY

The scientists directly involved in the work deny that speculation. Shi said her team did not find such antibodies, although she said some early tests did produce false positives that were corrected when the assays were fully validated. MIT Technology Review has been unable to locate Huang, but Gao said his lab never analyzed the miners’ antibody status, and that Huang’s statement—possibly based on the false-positive results, which Shi discussed at an internal meeting in 2012—was erroneous. After covid-19 struck, Shi’s team went back to the Mojiang samples to look for traces of SARS-CoV-2 proteins and found none. 

“Many pathogens can cause pneumonia-like symptoms similar to SARS and covid-19,” Zhong told me. Some local clinicians, he adds, still suspect it was a fungus that had sickened the miners. “It remains a mystery to this day.”

It’s not unusual for respiratory illnesses to have an unknown cause, but even though Shi couldn’t figure out what had sickened the Mojiang miners, her instinct told her that something interesting might be going on. “What viruses were lurking in the cave?” she remembers wondering. Between 2012 and 2015, her team undertook more than half a dozen trips to the mine shaft, about 1,100 miles from Wuhan, and collected 1,322 bat samples.

They looked for the coronavirus-specific RdRp gene, and when they found it, they investigated further. In the end, the bat samples turned out to contain nearly 300 coronaviruses. Nine belonged to the same group of viruses as SARS-CoV-1—known as beta-coronaviruses—even though their RdRp genes were quite different: they were “distant cousins,” Shi told me.

“Why are you so different?” Shi wondered, but eventually she put the sample back in the freezer.

Eight of the nine were closely related to each other, but one—from a single fecal sample labeled “4991”—had a very distinct genomic signature. “Why are you so different?” Shi wondered, but eventually she put the sample back in the freezer. Her work was to look for bat viruses that could potentially cause SARS-like epidemics, and none of the Mojiang sequences appeared to be “directly relevant to our inquiry,” she told me. “So they were not the focus of our research.”  

In 2018, though, 4991 was brought back out again. The Wuhan Institute of Virology had bought a new desktop sequencing machine, which made it much faster and cheaper to get a complete view of a virus’s genomic secrets, and 4991 was among the first batch of samples to be sequenced with the new device. The analysis confirmed that the virus residing in the sample was very different from SARS-CoV-1; they are 80% identical to each other across the genome. (The genomes of the other eight Mojiang viruses, which were sequenced after the pandemic, show they are only about three quarters identical to both SARS and covid-19 viruses across the genome.) It was always interesting to find new viruses, but there didn’t seem to be anything special for the researchers to write up, Shi said: “It didn’t seem to be a remarkable virus.” 

It was so unremarkable, in fact, that it was expendable: In their attempts to piece together its genomic makeup, the scientists used up all the sample. By 2018 the virus existed only as a sequence in the Wuhan institute’s database. 

In most cases, that would be the end of the story: the obscure, irrelevant virus would fade into oblivion. Except that it didn’t.

“I didn’t want to screw up”

At 5:30 in the morning on January 2, 2020, Si Haorui, a student on Shi’s team, headed toward the institute to start his day’s work. It was cold, and the white cloud of his breath danced around as he walked on the dark, empty street. 

Si is not a morning person. He rarely emerges before 10:30. But on that frigid January morning, he had a battle to fight. Two and half days earlier, clinicians at Wuhan Jinyintan Hospital, the city’s infectious-disease center, had sent samples to the virology institute for urgent analysis.

They were from seven patients in serious condition who had been recently hospitalized for a mysterious pneumonia.

The following day, December 31, the Wuhan Municipal Health Commission issued its first public statement about the outbreak, saying it was probing the cause of 27 pneumonia cases. Shi’s lab was among the first to officially investigate the illness, and Si was part of the team racing to pinpoint the cause. Working around the clock, team members had found coronavirus RdRp genes in five out of the seven patients’ samples; their next step was to sequence the viral genome. “That’s my specialty,” said Si, a slim man in his mid-20s whose eyes curve into two arcs when he smiles, the day we met at the institute’s sequencing facility. “I knew the stakes were high. I didn’t want to screw up.” 

(Shi’s lab was one of the four teams designated by China’s National Health Commission to work in parallel to pin down the cause of the new disease. This was a high-profile assignment, and only the commission had the authority to declare outbreaks of an emerging infectious disease and to release the relevant information.)

Stepping into the sequencing room felt like being a soldier stepping onto the battlefield, Si recalled. He had laid out his weapons the night before—the software he had tweaked for piecing together the genomic sequence of unknown pathogens. The machine was still running, busy reading short fragments of the genetic material from the bugs in those patients’ samples. The low humming sound of the machine filled the room. Si’s eyes were fixated on the sequencer. It reached the final stage of sequencing. It began processing the files. It took forever. Time seemed to stand still. Eventually it was done, and with a slightly shaky hand, he inserted a flash disk and copied the files over. He bolted upstairs to his office, where he could link to the institute’s supercomputer for the analysis. 

By 8:30 a.m., the genomic makeup had emerged. One sequence, now known as WIV04, was almost complete and of high quality: it was a coronavirus. 

Shi entered the sequence into the institutional and international databases to see if it was new. The closest match was the sequence from sample 4991, which the team had taken from Mojiang in 2013. The virus, no longer obscure or irrelevant, now deserved an official name. The team called it RaTG13—Ra for the bat species it was found in, Rhinolophus affinis; TG for Tongguan, the town where it was found; and 13 for the year of its discovery. It was, as they reported in Nature a month later, 96% identical to the coronavirus found in the new patients.

The fact that RaTG13 is so similar to SARS-CoV-2 has aroused suspicion. Critics like Alina Chan—a molecular biologist specializing in gene therapy at the Broad Institute of MIT and Harvard University in Cambridge, Massachusetts—wonder why Shi’s Nature paper published in February 2020 didn’t mention that RaTG13 came from the Mojiang mine where people had come down with the mysterious pneumonia. Chan, who leans strongly towards the lab leak theory, has helped it spread far and wide, and signed the Science letter calling for further investigation of the possibility. She said in Viral, a book she co-authored with the British science writer Matt Ridley, that the Wuhan institute had been “economical with the truth” about this. 

Shi attempted to head off this kind of suspicion by publishing an addendum detailing her Mojiang studies in Nature in November 2020 to show that the team had not detected any sign of coronavirus infection in the miners’ samples. But that didn’t help squelch the speculation.

The overall similarity between the two viruses, however, is not evidence that RaTG13 is the source of covid-19, according to an article published in Cell last September, authored by two dozen or so leading virologists and infectious-disease experts. The two viruses may be related, but they sit on different evolutionary branches that diverged half a century ago, says David Robertson, a virologist at the University of Glasgow in the UK. “RaTG13 couldn’t have naturally morphed into SARS-CoV-2,” he says. Neither could anybody have used RaTG13 as the backbone to engineer SARS-CoV-2, as some proponents of the lab leak theory have argued: the two viruses are different in 1,100 or so nucleotides spread across their whole genomes—a gap too large for any realistic effort. Making SARS-CoV-2 from RaTG13, says virologist Angela Rasmussen of the University of Saskatchewan in Canada, “would have required a feat of unprecedented genetic engineering.”

Meanwhile, evidence for the natural origins theory continues to mount. In the past year, several teams independent of the Wuhan institute have uncovered more than a dozen close relatives of SARS-CoV-2 in China, Japan, Laos, Thailand, and Cambodia. In a preprint paper posted in September 2021, a team of Laotian and French scientists reported the discovery of viruses in Laos that, according to Robertson, shared a common ancestor with SARS-CoV-2 as recently as a decade ago. These new discoveries are evidence that SARS-CoV-2 most likely evolved in the wild, says Robertson, who was not involved in the study. “We are closing in on the SARS-CoV-2 progenitor,” he says. 

But even if none of the bat coronavirus samples from Shi’s team are to blame for the pandemic, they aren’t the only viruses the scientists work with. Part of their research involves studying how the machinery of viruses works; and that has involved genetic mixing and matching of different pathogens to probe the function of viral genes. Could one of those chimeric viruses have been the source of the pandemic? To find out, I needed to talk to Shi.

Genetic tinkering

Bat woman takes her nickname seriously. A bat key ring lay on the desk in her office when I visited. A picture of her releasing a bat during a virus-hunting expedition hung near the window. Above the door was a green and yellow ceramic plate depicting a flying bat, which Shi bought on a field trip in Sichuan province. 

“Bats are a symbol of blessing in traditional Chinese culture,” she told me. They are called bian fu, meaning “flat” and “blessing,” respectively. “We often see bat motifs in jewelry, ceramics, and buildings in remote villages,” she said.

As the researchers’ collection of bat coronavirus sequences grew—especially after 2012, when they first managed to culture live viruses—they wanted to pinpoint the genetic ingredients that allow those viruses to infect humans, so scientists could develop drugs and vaccines to counter them. 

Shi was particularly interested in whether the spike protein was the sole factor that affected a coronavirus’s ability to infect cells, or whether other parts of the pathogen’s genome also had a role. One of her bat coronavirus sequences, SHC014, seemed ideal for such an inquiry. It was 95% identical to SARS-CoV-1 across the genome, but its spike was very different, and pseudovirus studies showed it was unable to facilitate entry into cells from several species, including humans. Did this mean that it was unable to infect humans? 

Scientists could not test this question directly because they hadn’t managed to isolate a live virus from the bat sample. But two genetic approaches could help shed light. One was to synthesize the virus from its genomic sequence; the other was to see whether SARS-CoV-1 could still cause disease if its spike was replaced with that of SHC014.

Shi didn’t have the necessary tools to do this type of genetic work, so in July 2013 she emailed Ralph Baric—a towering figure in viral genetics at the University of North Carolina at Chapel Hill—about joining forces along those lines of inquiry. 

The collaboration with Baric was not a close one, Shi told me: there was no exchange of laboratory staff, and Shi’s main contribution was to provide SHC014’s genomic sequence, which was yet to be published at the time. The findings, published in Nature Medicine in 2015, were surprising. It turned out that both the synthesized SHC014 and the SARS-CoV-1-SHC014 chimera were able to infect human cells and make mice sick. Both were less lethal than SARS-CoV-1, but—worryingly—existing drugs and vaccines that worked against SARS were unable to counter their effects. 

Meanwhile, Shi’s team was attempting similar tinkering in her own lab in a project funded by the US National Institutes of Health, which aimed to probe the genetic ingredients that could allow bat viruses to cause SARS-like diseases in humans. But while Baric focused on the human pathogen SARS-CoV-1 in the Nature Medicine paper, Shi used only its bat relatives—mostly WIV1, the first bat coronavirus the team had isolated. Their real-world risk to humans was unknown. By the time the pandemic broke out, her team had created a total of a dozen or so chimeric viruses by swapping WIV1’s spike with its counterpart from newly identified sequences of bat coronaviruses, only a handful of which could infect human cells in a petri dish. 

There were more surprises in store. In an unpublished experiment, released by the NIH in response to a Freedom of Information Act lawsuit brought by The Intercept, the researchers tested the ability of three such chimeras to infect mice expressing human ACE2. Compared with their parental strain, WIV1, the three chimeric viruses grew a lot more quickly in the mouse lungs in the early stage of the infection, but WIV1 caught up by the end of the experiment. 

The differences surprised Shi, but what puzzled her the most was that the chimera causing the most weight loss in infected mice—an indicator of its pathogenicity—was WIV1-SHC014, whose spike was most dissimilar to that of SARS-CoV-1. The one whose spike was most similar had no effect on the animals’ weight.

The results from genetic studies in both Baric’s and Shi’s labs—both collaborating with the New York–based EcoHealth Alliance—have provided compelling evidence that the spike protein is not the only factor in whether a virus can make an animal sick, researchers say. “We can’t assess the emergence potential of viruses using only pseudovirus assays or predictions based on genomic sequences and molecular modeling,” Shi told me. 

None of the chimeras created in Shi’s labs was closely related to SARS-CoV-2, and therefore, none could have been the cause of the pandemic. But it does seem that the team created at least one chimeric virus, WIV1-SHC014, with a functional gain—that is, increased pathogenicity—relative to the parental strain, WIV. Critics like Richard Ebright, a molecular biologist at Rutgers University, regard this as the type of gain-of-function research that ought to be subject to strict regulatory oversight. But Shi says that in none of those studies—including her collaborations with Baric and with EcoHealth—did the teams intend to create more dangerous viruses. None of the chimeras had been reasonably anticipated at the time of proposal to have increased transmissibility or pathogenicity in mammals. 

According to an NIH spokesperson, the grant Shi jointly applied for with the EcoHealth Alliance—the only one with a sub-award to the Wuhan institute—“was reviewed and determined by experts to fall outside the scope” of its regulatory framework for gain-of-function research.

Virologists such as the University of Utah’s Goldstein argue that such genetic studies could help protect us from future pandemics. In the past year, research teams including Baric’s have demonstrated the possibility of developing so-called pan-coronavirus vaccines that could simultaneously block a group of coronaviruses—including SARS-CoV-1, SARS-CoV-2, their bat relatives that Shi has discovered, and potentially other relatives that are yet to be identified. Last September, NIH announced an award of $36.3 million to further such work. Discovering novel viruses in the wild and using genetic techniques to probe their function in the lab, researchers say, could point toward ways of mitigating and treating future disease outbreaks similar to SARS and covid-19. 

Biosafety challenges

Even though none of those chimeric viruses was the source of covid-19, there are still concerns that the biosafety standards in the Wuhan lab might not have been rigorous enough to prevent research activities from causing the pandemic.

Studies involving live viruses and genetic tinkering are inherently risky. Accidents can happen even with the most stringent biosafety precautions in place. Scientists might get inadvertently infected in the lab; genetic mixing and matching might unexpectedly create a superbug whose ability to escape overmatches the biosafety designation of its parental strains.

I asked Shi how China regulates coronavirus research to minimize the risks. 

“China doesn’t have a blanket biosafety regulation on all coronavirus research,” she said. “Everything is assessed on a case-by-case basis.” Studies of SARS-CoV-1 and SARS-CoV-2, for instance, have to be done in BSL-3 labs, whereas the human coronaviruses that cause the common cold are handled under BSL-2 conditions. What about bat viruses? 

The Wuhan institute’s biosafety committees ruled a decade ago that while work with animals must be carried out in BSL-3, molecular and cell-culture work involving bat coronaviruses can be done in BSL-2, albeit in biosafety cabinets with air filtration and under negative pressure to keep viruses inside.

Some scientists, like Ebright, regard this as unsafe. Bat coronaviruses are, as he puts it, “uncharacterized agents” with unknown virulence and transmissibility. “The only acceptable approach is to start with a high biosafety-level assignment … and to lower the biosafety-level assignment only if and when it is determined it is prudent to do so,” he told me in an email. 

Others, however, don’t think Shi’s work indicates lax biosafety standards in China. The dominant view among scientists worldwide was—and to some extent still is—that bat coronaviruses would most likely have to evolve in an intermediate animal first before they could infect humans. “Every institute’s biosafety committee has to balance the real risks with the potential risks,” says the University of Saskatchewan’s Rasmussen, adding that the Wuhan institute’s biosafety designation was reasonable at the time. 

And it’s not uncommon for labs around the world to culture uncharacterized animal viruses in BSL-2 facilities. Ebright told me in an email that current US guidelines place only three coronaviruses—SARS-CoV-1, SARS-CoV-2, and MERS-CoV—under BSL-3 rules. Some contagious animal coronaviruses that can infect human cells in a petri dish, including deadly pig viruses that originated in bats, are—like Shi’s viruses—designated BSL-2 agents. (In the US, culturing rabies virus, another deadly pathogen that often lives in bats, is also designated as a BSL-2 task even though the virus has a fatality rate of nearly 100% in humans.)

Rasmussen told me that the emergence of covid-19 means we should reevaluate those biosafety standards for viruses with unknown risks. “I think the pandemic has changed that risk-benefit equation,” she said. 

China’s high-level laboratories face other challenges besides the difficulty of making biosafety judgment calls. Money is one major issue. While there’s often ample funding to purchase cutting-edge equipment and build state-of-the-art laboratories such as the Wuhan institute’s BSL-4 facility, scientists often struggle for funding to train workers or to cover the cost of running those labs. 

Such obstacles are hardly a secret. When the US embassy in Beijing sent a delegation to visit the Wuhan Institute of Virology in early 2018, managers of the institute lamented about them to embassy staff. And Yuan Zhiming, director of the BSL-4 facility, detailed the challenges of high-level biosafety laboratories in China in a paper in September 2019.

Some have painted such challenges as a clear sign of lax standards. In an article published in April 2020, Washington Post columnist Josh Rogin wrote that after the US officials’ visit of the Wuhan institute in 2018 they “sent two official warnings back to Washington about inadequate safety at the lab.” According to Rogin, unnamed sources familiar with the unclassified cables “said that they were meant to sound an alarm about the grave safety concerns,” and one anonymous Trump administration official told him the cables “provide one more piece of evidence to support the possibility that the pandemic is the result of a lab accident in Wuhan.” 

The newspaper column marked a turning point in the debate over covid-19’s origins, catapulting the lab leak theory into the mainstream. Several mainstream media outlets have used its assertions as evidence that the Wuhan institute has a record of “spotty” or “shoddy” biosafety practice.

The cables themselves, which were publicly released several months later (with some parts redacted), cautioned about inadequate staffing but didn’t identify any specific dangerous biosafety practices. One cable, sent on January 19, 2018, mentioned the shortage of trained staff “needed to safely operate this high-containment laboratory” in a section that discussed how a lack of trained workers could “impede research.” According to the second cable sent three months later, this “opens up even more opportunities for expert exchange.” The January cable also noted the Wuhan institute’s ability “to undertake productive research despite limitations” and said that the work “makes the continued surveillance of SARS-like coronaviruses in bats and study of the animal-human interface critical to future emerging coronavirus outbreak prediction and prevention.”

Some scientists are appalled by what they perceive as misrepresentation of the embassy cables. “The concerns raised in the cable did not appear to focus on any specific safety concerns or egregious activities within the laboratory by current staff,” Jason Kindrachuk, an infectious-disease expert at the University of Manitoba in Winnipeg, Canada, told me in an email. It highlighted, he adds, how “these current limitations may be remedied through” additional help from the international community, including the US. In any case, Bill Hanage, an infectious-disease expert at Harvard, told me in an email that he doesn’t think the existence of the cables shed any light on the covid-19 origins debate. 

Rogin told MIT Technology Review in an email that he stands by his reporting in his 2020 article.

Shi says that the lack of trained staff means that China cannot make the most out of the facility, but it doesn’t mean that it was using untrained personnel to work in BSL-3 or BSL-4 labs. The Wuhan institute, she adds, abides by the international norms of biosafety governance and that her research before the pandemic was geared toward bat viruses closely related to the original SARS virus. “RaTG13 was the closest SARS-CoV-2 relative we had ever had,” she said. “We could not have leaked what we did not have.”

Shi also denied suggestions that the first human infection could have involved someone from her team—who caught the virus either in the lab or in the field. Between the beginning of the outbreak in Wuhan and the first vaccine shots, she told me, every member of her team was tested multiple times for viral nucleic acids to detect ongoing infections and for antibodies that would indicate past exposure. “Nobody was tested positive,” she said. “None of us has been infected by coronaviruses under any circumstances, including while sampling bats in the field.”

Politics of mistrust

Many scientists are dismayed by the way Shi and the Wuhan Institute of Virology are often portrayed in Western media. Even those with no connection to Shi or the Wuhan institute—such as the University of Glasgow’s Robertson and the University of Saskatchewan’s Rasmussen—call it shockingly biased and say it is driven partly by geopolitical motives and deep-rooted prejudice.

To China experts like Joy Zhang, a sociologist at the University of Kent in Canterbury, UK, who specializes in science governance in China, it’s hard to separate the specific allegations against Shi from the general suspicions of China. “Shi is a victim of the Western mistrust of China and Chinese science,” she says. 

Such mistrust of Chinese scientific practices is obvious among some. Filippa Lentzos, a biosecurity expert at King’s College London, told me in February last year that “it’s simply too late” to find out what happened because “everything, for instance, in the Wuhan Institute of Virology freezers would have been cleared out. The data records would have been scrubbed or cleaned up.” She says it’s still her view now.

Shi finds Lentzos’s allegations that her lab would destroy critical records “baseless and appalling.” 

“If that’s what they think, then there is nothing we can do to convince them otherwise,” she told me. “Even if we gave them all the records, they would still say we have hidden something or we have destroyed the evidence.”

Some in the West agree. “I’m quite distressed by people throwing this kind of extremely serious allegation around,” Nancy Connell, a microbiologist and member of NIH’s National Science Advisory Board for Biosecurity, told me in February last year, when she was with the Johns Hopkins Center for Health Security. “It’s highly irresponsible.” 

But even if the lab leak theory is partly fueled by a deeply rooted mistrust of China, the country’s questionable credibility record and a sequence of curious missteps have not helped. 

During the SARS outbreak in 2002-’03, Chinese officials downplayed its extent for months until a prominent military surgeon blew the whistle. At the onset of covid-19, China also obscured information about the early cases and clamped down on domestic debate. This was exacerbated when, in March 2020, a number of Chinese ministries ruled that scientists had to seek approval to publish any work related to covid-19 research. 

Meanwhile, several Chinese institutions, including the Wuhan Institute of Virology, instructed their scientists—with rare exceptions—not to speak to the press. For some, this was something of a relief. Conducting interviews on politically sensitive subjects in English is prohibitively daunting to many Chinese speakers, as any language errors, especially regarding tenses and auxiliary verbs, can easily be misconstrued—with grave consequences. At the same time, many Chinese scientists had become reluctant to talk to Western journalists for more straightforward reasons: the majority of reporters who had contacted them, they said, didn’t seem to understand the intricacies of the science and showed strong preconceived ideas. 

“I just wanted to put my head down and concentrate on my work,” Shi told me. “I thought the storm would just blow over after some time.” 

Some of the Wuhan institute’s behavior has certainly raised red flags. In February 2020, for example, it took its virus databases offline, and they remain unavailable to outsiders—prompting some to suggest that they might contain information critical to covid-19’s origins. Shi told me that the part of the databases that had been publicly available before the pandemic contained only published information; the Wuhan institute, like research organizations in other parts of the world, had unpublished data that could be shared upon request via portals for academic collaborations. The institute, she says, took the databases offline because of security concerns; there had been thousands of hacking attempts since the beginning of the pandemic. “The IT managers were really worried somebody might sabotage the databases or, worse, implant virus sequences for malicious intent,” she said.

Instead of tackling the publicity crisis directly, China has exacerbated mistrust by running obfuscation and disinformation campaigns of its own.

Still, the University of Kent’s Zhang says, China’s behavior has to be understood in the country’s larger political, media, and cultural context. China, with its totally different media tradition, “has neither the vocabulary nor the grammar of the Western press to deal with a publicity crisis,” she told me. “The first instinct of Chinese officials is always to shut down communication channels.” To them, she said, this often seems safer than dealing with the situation proactively. Several top Chinese scientists, who asked not to be named for fear of political repercussions, told me that this also reflects a lack of confidence among China’s top leaders. “While eager to assert itself as a global power, China is still terribly insecure,” one of them said.

Instead of tackling the publicity crisis directly, China has exacerbated mistrust by running obfuscation and disinformation campaigns of its own. Its foreign ministry, for instance, has insinuated that biomedical labs at a military base in Maryland may have created SARS-CoV-2 and leaked it to the public. Then there are the apparent falsehoods. The Chinese members of the WHO mission insisted in their report that “no verified reports of live mammals being sold [at the Huanan market] around 2019 were found.” In June, however, a paper published in Scientific Reports showed that many vendors sold live mammals illegally at several markets in Wuhan, including the Huanan market, just before the pandemic. 

Many scientists in the West are dismayed by such obfuscation. Even those who consider the lab leak theory highly unlikely are adamant that this behavior is unacceptable. “If China is lying about this, what else is it lying about?” says one virologist who strongly supports the natural origins theory. 

Wu Zhiqiang—a virologist with the Beijing-based Institute of Pathogen Biology at the Chinese Academy of Medical Sciences and a member of the WHO mission—denies that his team lied. He told me that tracking down illegal wildlife trade was beyond the scope of the scientific mission. “We had to work with the information provided by the various ministries and were unable to verify the sale of live mammals at the Huanan market,” he says. Studies of disease origins, he adds, are always based on incomplete data, but Chinese scientists are following up clues to probe the market link: “It takes time and patience to learn the scientific truth.”

Adding fuel to the mistrust, though, is the role of the EcoHealth Alliance’s Daszak. His close ties with Shi’s lab and his role as a member of the WHO mission’s international team are potentially in conflict. Critics say he can also be less than forthcoming. In February, for instance, he told several media outlets that he was impressed with China’s openness—at a time when the team was under tremendous pressure to conform to the Chinese narrative. While giving the impression that he knows very well what’s going on at the Wuhan institute, Daszak and his organization have also provided incorrect statements about its research activities.

Such incidents, critics say, have raised questions about whether Daszak had a disproportionate—or even misleading—role in the WHO mission. But scientists like the University of Utah’s Goldstein, who do not collaborate with Daszak, told me that there is no evidence that Daszak “wielded disproportionate influence” in the 11-member team.

Daszak told me in an email that his potential conflicts of interest had been declared to the WHO before he joined the mission team. He says that there is lots of misreporting about him and his work in the media and that he is often not given the chance to respond to accusations. EcoHealth Alliance, he adds, has acted “with scientific integrity and honesty.” 

“It’s now over two years since the first efforts to willfully politicize the pandemic origins, and to undermine science and the work that scientists do in often difficult circumstances,” says Daszak. “All of us have lost due to this politicization. When you mix politics with science, you get politics.”

“Clear and immediate threat”

On a hot July afternoon last year, I joined Shi and her team on a virus-hunting trip to a bat cave in Hubei province. (The team does not want the exact location of the cave disclosed, to avoid unwelcome media attention.) Dusk was falling fast, and the air smelled acrid and musty. Thousands of horseshoe bats clung to the cave ceiling—quiet, motionless, and evenly spaced out, like fighter jets on an airfield waiting for orders to take off. 

To capture bats, researchers used a gigantic net made of fine nylon mesh suspended between two poles. Shi and Yang pushed the poles against the entrance of the cave, adjusting their position to cover the gaps between the net and the rocks. We switched off our headlamps and waited in the dark. Moments later, a fluttering sound ricocheted above us. A shadow swirled around and shot into the net, like insects flying into a spider web. The bat immediately got tangled. “Here we go,” shouted Shi. “Our first catch!” 

The cave, at the bottom of a lush hill in a small village, is Shi’s home base. She uses it for sampling viruses, training students, and developing technologies that trace the movements of bats and the pathogens they carry. So far, it has yielded only distant relatives of known coronaviruses; their significance is unclear. (Bats in another cave in Hubei, however, have yielded SARS-like viruses.) “We are just collecting pieces of the jigsaw puzzle,” Shi told me. “We never know what will cause the next pandemic.” 

And the team keeps doing that work. The pandemic has lent extra urgency to one aspect of its research: determining the exposure risks that rural people face. In previous studies, Shi and her colleagues found that up to 4% of people living close to bats and working closely with wildlife in southern China were infected with dangerous animal-borne viruses, including coronaviruses; the infection rate was 9% among butchers. The Laotian and French team that discovered close relatives of SARS-CoV-2 found that one in five people who’d had direct contact with bats and other wildlife had coronavirus antibodies. 

Such findings suggest that viruses closely related to SARS-CoV-2 might be spreading over a massive geographic range, stretching at least 3,000 miles from Japan to Cambodia. A combination of population growth, wildlife trade, rampant deforestation, and improved transportation in those places has made it increasingly easily for animal pathogens to cross over to humans. 

Robertson, the University of Glasgow virologist, says this is a clear and immediate threat: “It’s quite terrifying, really, to think how we can fuck this up by not finding out where [those viruses] are and risk more spillover.”

To watch out for viruses jumping between species, many scientists say, China should build on the WHO mission findings and set up long-term surveillance. Perhaps farms in southern China that supplied animals to the Huanan market should be a focus, or species known to be susceptible to SARS-CoV-2, such as civets, minks, badgers, raccoon dogs, and people who live close to wildlife or work in the animal trade. This wouldn’t just help pin down the origins of covid-19, says Fabian Leendertz, an expert on zoonotic diseases and the founding director of the Helmholtz Institute for One Health in Greifswald, Germany, who was a member of the WHO mission. “It’s also about reducing the risk of the next pandemic,” he says . “It can help strengthen capacity building in neglected rural areas. It should be a concerted global effort.”

But such international collaborations with China are getting increasingly unlikely because of the allegations leveled at the Wuhan institute. 

Meanwhile, according to a WHO spokesperson, all hypotheses are still on the table and the lab leak theory would require further investigation, potentially with additional missions involving biosafety and biosecurity experts. Last November, the WHO put together an advisory group to probe the origins of covid-19 and future epidemics and to guide studies of emerging pathogens. The group, says the spokesperson, will release its first set of recommendations in the coming weeks. 

Shi now realizes the controversial nature of her work and agrees that there’s an urgent need to step up regulation and oversight of risky research. She welcomes a broader societal debate about searching for new viruses in the wild and tampering with their genomes in the lab—which some biosafety experts ardently oppose. But “they don’t have to crucify me for that end,” she told me.

After talking to dozens of scientists involved over the past year, it has become clear to me that people’s opinions about the lab leak theory, to a large extent, depend on whether or not they believe Shi. Some support her, partly because they know her as a person or understand her work, or because they are willing to put up with the messy reality of science and China’s lack of transparency. Others, possibly driven by a deep mistrust of China, grave biosafety concerns, or an intense desire for greater transparency, simply reject every piece of evidence that she offers to define her work, and regard any inconsistencies as deliberate attempts to cover up a crime. 

Not surprisingly, the allegations have taken a personal toll. “I’m a human being as well, you know,” Shi told me. “Have they considered what it feels like to be wrongly accused of unleashing a pandemic that has killed millions?” 

Since the outbreaks, Shi has received numerous abusive emails and phone calls, even death threats. She has been called a liar, a mass murderer, and an accomplice of the Chinese Communist Party (even though she’s not a member). In May 2020, it was falsely rumored that she had defected to France with nearly 1,000 classified documents. 

At Shi’s bat-themed office, I asked her how the past two years have marked her. Her girlish face suddenly dimmed. 

“I can’t bear looking back,” she said, and turned her head away. 

A long silence ensued. 

“I used to admire the West. I used to think it was a just and meritocratic society. I used to think it must be wonderful to live in a country where anybody could criticize the government.”

“What do you think now?” I pressed.

“Now I think if you are Chinese then it doesn’t matter how good you are at your job—because you are tried by nationality,” she said. “I’ve now realized that the Western democracy is hypocritical, and that much of its media is driven by lies, prejudices, and politics.” 

Shi paused and drew a sharp breath. Her body tensed, blood flushing her cheeks. The air swelled and seemed to grow hotter.

“They’ve lost the moral high ground as far as I’m concerned,” she said. And if politics overpowers science, “then there will be no basis for any cooperation.” 

The reporting was supported by a grant from the Pulitzer Center.

 Jane Qiu is an award-winning independent science writer based in Beijing and a former Knight Science Journalism Fellow at MIT.

The metaverse, which graduated from a niche term to a household name in less than a year, is an excellent case in point. Its metamorphosis began in July 2021, when Facebook announced that it would dedicate the next decade to bringing the metaverse to life. In the company’s presentation of the concept, the metaverse was a thing of wonder: an immersive, rich digital world combining aspects of social media, online gaming, and augmented and virtual reality. “The defining quality of the metaverse will be a feeling of presence—like you are right there with another person or in another place,” Facebook founder Mark Zuckerberg wrote, envisioning a creation that would “reach a billion people, host hundreds of billions of dollars of digital commerce, and support jobs for millions of creators and developers.” By December 2021, a range of other large American technology companies, including Microsoft, Intel, and Qualcomm, had all articulated metaverse plans of their own. And by the time the Consumer Electronics Show rolled around in January, everyone seemed to have a metaverse angle, no matter how improbable or banal: haptic vests, including one with an air conditioner to create your own localized climate; avatar beauty makeovers; virtual delivery vans for your virtual home. 

There has been plenty of discussion about the involvement of Meta (née Facebook) and its current complicated position as a social media platform with considerable purchase on our daily lives. There have also been broader conversations about what form the metaverse could or should take, in terms of technical capabilities, user experiences, business models, access, and regulation, and—more quietly—about what purpose it would serve and what needs it would fulfill.

“There is an easy seductiveness to stories that cast a technology as brand-new.”

These are good conversations to have. But we would be remiss if we didn’t take a step back to ask, not what the metaverse is or who will make it, but where it comes from—both in a literal sense and also in the ideas it embodies. Who invented it, if it was indeed invented? And what about earlier constructed, imagined, augmented, or virtual worlds? What can they tell us about how to enact the metaverse now, about its perils and its possibilities? 

There is an easy seductiveness to stories that cast a technology as brand-new, or at the very least that don’t belabor long, complicated histories. Seen this way, the future is a space of reinvention and possibility, rather than something intimately connected to our present and our past. But histories are more than just backstories. They are backbones and blueprints and maps to territories that have already been traversed. Knowing the history of a technology, or the ideas it embodies, can provide better questions, reveal potential pitfalls and lessons already learned, and open a window onto the lives of those who learned them. The metaverse—which is not nearly as new as it looks—is no exception. 

So where does the metaverse come from? A common answer—the clear and tidy one—is that it comes from Neal Stephenson’s 1992 science fiction novel Snow Crash, which describes a computer-generated virtual world made possible by software and a worldwide fiber-optic network. In the book’s 21st-century Los Angeles, the world is messy, replete with social inequities, sexism, racism, gated communities, surveillance, hypercapitalism, febrile megacorporations, and corrupt policing. Of course, the novel’s Metaverse is messy too. It too heaves with social inequities and hypercapitalism. Not everyone finds their way there. For those who do, the quality of their experience is determined by the caliber of their kit and their ability to afford bandwidth, electricity, and computational horsepower. Those with means can have elaborately personalized digital renderings. Others must make do with simple flat sketches, purchased off the shelf—the “Brandy” and “Clint” packages. Perhaps we shouldn’t be surprised that many who read the book saw it not just as cutting-edge science fiction but as a critique of end-stage capitalism and techno-utopian visions.

In the three decades that have passed since Snow Crash was published, many of the underpinnings of Stephenson’s virtual world, such as social networks and artificial intelligence, have materialized. And the metaverse, like other ideas foreshadowed in the cyberpunk tradition, has persistently found its way into broader conversation. It has featured in recent movies such as Ready Player One and Free Guy. And it has shaped much of the digital landscape in which we now find ourselves. However, I think there might be more to the metaverse than just Snow Crash and its current re-instantiation.

In fact, today’s conversations around the metaverse remind me a lot of the conversations we were having nearly 20 years ago about Second Life, which Philip Rosedale’s Linden Lab launched in 2003. Rosedale is very clear about the ways in which he was inspired by Snow Crash. He is also clear, however, that a trip to Burning Man in the late 1990s forever framed his thinking about virtual worlds, their inhabitants, and their ethos. Second Life was to be “a 3D online world created and owned by its users.” It was hugely successful—it dominated news headlines and conversations. Companies and brands fought to establish themselves in this new domain; we had conferences and concerts in Second Life, and even church. In the early 2000s, millions of people flocked to the platform and created lives there. Anthropologists studied them*; policy makers and politicians debated them. And the realities of a fully fledged virtual world collided quickly with regulators and policy makers; concerns about fiat currencies, money laundering, and prostitution all surfaced. 

However, I think there are even earlier histories that could inform our thinking. Before Second Life. Before virtual and augmented reality. Before the web and the internet. Before mobile phones and personal computers. Before television, and radio, and movies. Before any of that, an enormous iron and glass building arose in London’s Hyde Park. It was the summer of 1851, and the future was on display. 

Arc lights and hydraulic presses (powered by a hidden steam engine), electric telegrams, a prototype fax machine, mechanical birds in artificial trees, a submarine, guns, the first life-size and lifelike sculptures of dinosaurs, Goodyear’s vulcanized rubber, Matthew Brady’s daguerreotypes, even Britain’s first flushing public toilets. There were three stories’ worth of alcoves with red bunting and signs proclaiming each display’s country of origin, spread out over 92,000 square meters of gleaming glass enclosures—the Crystal Palace, as one satirical magazine dubbed it.

It was a whole world dedicated to the future: a world in which almost anyone could be immersed, educated, challenged, inspired, titillated, or provoked. 

The Great Exhibition of the Works of Industry of All Nations, as the extraordinary event was formally known, was the brainchild of Prince Albert, Queen Victoria’s beloved consort. It would showcase more than 100,000 exhibits from all over the world. The queen herself would attend at least 30 times. In her opening speech, she made clear her agenda: “It is my anxious desire to promote among nations the cultivation of all those arts which are fostered by peace and which in their turn contribute to maintain the peace of the world.” The age of empire may already have been in decline, but the Great Exhibition was all about asserting power and a vision for Britain’s future. And what a modern, industrialized future it would be, even if colonies all over the world would be needed to make it happen. 

Of course, London was a city already full of expositions and displays, places where you could visit the wondrous and strange. Charles Babbage was partial to Merlin’s Mechanical Museum, with its many automata. Others favored dioramas of the Holy Land and Paris. The Great Exhibition was different because it had scale, and the power of empire behind it. It wasn’t just a spectacle; it was a whole world dedicated to the future: a world in which almost anyone could be immersed, educated, challenged, inspired, titillated, or provoked. It was not little bits and pieces, but one large, imposing, unavoidable statement. 

In its day, the Great Exhibition had many critics. Some worried about the ancient elm trees in Hyde Park that found themselves contained in the enormous structure. Others worried about the tensile strength of all that glass. In the press, there were months of ridicule, with one politician describing it as “one of the greatest humbugs, frauds, and absurdities ever known.” In the Houses of Parliament, some questioned Prince Albert’s motives, citing his status as a foreign prince and suggesting that the Great Exhibition was just a publicity exercise to encourage and perhaps mask the rise of immigration in Britain. Still others suggested that the Great Exhibition would attract pickpockets, prostitutes, and spies, and called for 1,000 extra police to be on duty. 

Unsurprisingly, the dire warnings were overblown, and for a sunny summer, people from all over Britain—taking advantage of the rapidly expanding railway network—flocked to the massive glass house in the park. The organizers set entrance fees at a shilling, which made it accessible to the British working classes. “See the world for a shilling” was a common refrain that summer. 

A surprising fraction of the literary and scientific community of the day found its way to the Crystal Palace. That roll call includes Charles Dickens, Charles Dodgson (who would become Lewis Carroll), Charles Darwin, Karl Marx, Michael Faraday, Samuel Colt, Charlotte Brontë, Charles Babbage, and George Eliot. Dickens hated it: it was just all too much rampant materialism, and his most recent biographer claims that his experiences there shaped all his work thereafter. Brontë, by contrast, wrote, “It seems as if only magic could have gathered this mass of wealth from all the ends of the earth—as if none but supernatural hands could have arranged it thus, with such a blaze and contrast of colours and marvelous power of effect.” Dodgson had such a moment when he entered the Crystal Palace. He wrote, “The impression when you get inside is of bewilderment. It looks like a sort of fairyland.”

And then, just like that, the Great Exhibition closed its doors on the 15th of October, 1851. Over its five-and-a-half-month run, it was estimated, over 6 million people visited the Crystal Palace (at the time, the total population of Britain was only 24 million). In its short life in Hyde Park, the Great Exhibition also turned a remarkable profit of some £186,437 (more than $35 million today). Some of it went to the purchase of land in South Kensington to create London’s current museum district. Another portion underwrote an educational trust that still provides scholarships for scientific research. The Crystal Palace was disassembled in the winter of 1851 and transported to a new site, where it would continue to showcase all manner of wonders until a cataclysmic fire in 1936 reduced it to a smoldering iron skeleton. And if the fancy takes you, you can still visit the Great Exhibition today, via a virtual tour hosted on the website of the Royal Parks. 

The Great Exhibition kicked off more than a century of world’s fairs—spaces of spectacle and wonder that, in turn, would shape the world around them. In America, these world-making activities included the World’s Columbian Exposition of 1893, also known as the Chicago World’s Fair—a whole city with more than 200 purpose-built structures, whitewashed and gleaming, showcasing technologies as varied as a fully electrical kitchen with dishwasher, an electric chicken incubator, a seismograph, Thomas Edison’s kinetoscope, searchlights, Morse code telegraphy, multiphase power generators, moving walkways, and the world’s first Ferris wheel. Over one quarter of Americans would attend the World’s Fair in less than six months.

If the Great Exhibition had celebrated the power of steam, this so-called White City was all about electricity. It was also a branded landscape, supported and then aggressively promoted by American industry, with soon-to-be-familiar names like General Electric, Western Electric, and Westinghouse showcasing their technologies and their visions for the future—American democracy and American capitalism. Complicated conversations about gender and racial equality, and mythologizing of American exceptionalism and individualism, were everywhere on display. There was, for example, a building dedicated to the lives and times of American women, but not one for African-Americans, a point fiercely argued by Ida B. Wells and Frederick Douglass, who saw an opportunity to celebrate African-American accomplishments since the Emancipation Proclamation.

The White City also ushered in a new type of spectacle. At the Midway Plaisance, a mile-long stretch of park on the edge of the exposition site, you could see people on display in living dioramas, intermixed with dedicated sideshow activities, amusements, concessions, and food stalls. It was a violent, exciting mess of orientalism, exclusion, appropriation, and celebration. And it was far and away the most popular destination in the White City, generating a significant profit—$4 million in 1893 dollars, or well over $100 million today. 

The Midway would in turn inspire the creation of Coney Island in New York, and ultimately California’s Disneyland—a wholly different brand of imagined world. The influence of these kinds of events on our imaginations should not be underestimated. Just as there is a straight line from the Midway to Coney Island to Disneyland, there is a straight line from the White City to the 1939 New York World’s Fair to the Consumer Electronics Show. We can also draw a line between the Great Exhibition and today’s metaverse. Like the virtual world that the metaverse’s promoters promise, the Great Exhibition was a world within the world, full of the splendors of its day and promises about the future. But even as it opened up new spaces of possibility—and profit—it also amplified and reproduced existing power structures through its choices of exhibits and exhibitors, its reliance on the Royal Society for curation, and its constant erasure of colonial reality. All this helped ensure that the future would look remarkably British. The exhibition harnessed the power of steam and telegraphy to bring visitors to a space of new experiences, while masking the impact of such technological might; engines and pipes were hidden underground out of plain sight. It was a deliberate sleight of hand. If Brontë saw magic—not power, xenophobia, and nationalism—that was what she was intended to see.

To support MIT Technology Review’s journalism, please consider becoming a subscriber.

I think our history with proto-­metaverses should make us more skeptical about any claims for the emancipatory power of technology and technology platforms. After all, each of them both encountered and reproduced various kinds of social inequities, even as they strove not to, and many created problems that their designers did not foresee. Yet this history should also let us be alive to the possibilities of wondrous, unexpected invention and innovation, and it should remind us that there will not be a singular experience of the metaverse. It will mean different things to different people, and may give rise to new ideas and ideologies. The Great Exhibition generated anxiety and wonder, and it alternately haunted and shaped a generation of thinkers and doers. I like to wonder who will author this metaverse’s Bleak House or Alice in Wonderland in response to what they encounter there. 

The Great Exhibition and its array of descendants speak to the long and complicated human history of world-making. Exploring these many histories and pre-­histories can be generative and revelatory. The metaverse will never be an end in itself. Rather, it will be many things: a space of exploration, a gateway, an inspiration, or even a refuge. Whatever it becomes, it will always be in dialogue with the world that has built it. The architects of the metaverse will need to have an eye to the world beyond the virtual. And in the 21st century, this will surely mean more than worrying about ancient elm trees and the tensile strength of glass. It will mean thinking deeply about our potential and our limitations as makers of new worlds.

Genevieve Bell is director of the School of Cybernetics at the Australian National University in Canberra.

* Two lovely ethnographic accounts of Second Life grace my shelves: Tom Boellstorff’s Coming of Age in Second Life: An Anthropologist Explores the Virtually Human (2008) and Thomas Malaby’s Making Virtual Worlds: Linden Lab and Second Life (2009). The former is an excellent account of the early years of Second Life and the ways in which people loved and loathed that virtual world; the latter focuses on the technologists who built Second Life. Both give insight into the utopian visions that underpinned Second Life, and how they were experienced by participants and builders alike.

It’s not only the rising drumbeat of the bad news that keeps this threat top of mind. When you regularly see the impacts on your industry peers, you start asking yourself: are we next? At the GDPI launch event, Michael Dell, chairman and CEO of Dell Technologies, explained why all businesses, large and small, from your insurance broker to the local butcher, are more spooked than ever before.

The GDPI survey uncovered that 64% of leaders are concerned they’ll experience a disruptive event, such as data loss or downtime, in the next year. With the frequency of ransomware attacks on the rise, all businesses should expect an attack. Whether or not you should be fearful depends on how prepared you are.

A threat like no other

Many cybersecurity threats are destructive, but few pack as big a punch as ransomware. Its profound effects stretch across your entire organization, halting operations, disrupting business-critical services, and sometimes even putting people at risk. These attacks are also among the costliest to mitigate.

What makes ransomware unique, however, is its “in your face” style. You can discreetly mitigate other security incidents, but ransomware attacks have become so overt that your customers will most likely know about them. What would that do to your brand reputation and trust?

The perfect crime

For cybercriminals, ransomware is the perfect crime for the digital age. Not only does it have a low entry barrier, but it yields a greater return on investment than garden-variety cybercrime. Like a savvy entrepreneur, a threat actor goes where the best opportunities are—and today, that’s ransomware.

A ransomware attack requires little technical skill, thanks to the availability of ransomware-as-a-service on the dark web marketplace. The ransomware operators don’t have to concern themselves with reconnaissance, gaining initial access or writing exploits. All these services, and plenty others, are available in abundance—complete with 24/7 customer service.

On top of that, the attackers don’t have to go far to monetize. When you’re hit with ransomware, you become, in essence, an instant “customer” of theirs. They know you need your systems to be up and running as fast as possible, and you need to prevent the potential release of your data. They have your instant attention and the power—unless you have the means to defend yourself and recover your data.

Defense starts with the basics

To guard against ransomware, you have to start with the basics. First, implement the NIST Cybersecurity Framework (or a similar framework designed for your industry). Once you have the essential pieces in place—patching, antivirus, security awareness, and so on—you can build to the more sophisticated defenses, such as zero-trust and identity and access management.

Regardless of what other defenses you have in place, one of the most critical steps in fighting a ransomware infection is data backup. The more robust your backup plan, the less power and hold the attackers will have over you.

So, what’s your backup plan?

You likely have a backup strategy, but have you considered how ransomware has evolved? Before compromising your core data, attackers will typically spend a little bit of extra time in your network to see if they can compromise your backups. If you have a connected backup, they’ll find a way to exploit it.

That’s why you need an immutable, offline copy for your critical systems. But if this immutable copy is at some distant location on tapes, how quickly can you access it and restore your systems? According to the GDPI survey, the average time to recover from disruption, such as a ransomware attack, is six hours. But that length of time is too disruptive for many organizations.

Founders Federal Credit Union (FFCU) calculated that they could only give themselves an hour window. Working in a high-volume, online transaction-based industry, they simply couldn’t afford more time. So, the financial institution implemented a major overhaul of its data center with a focus on cyber resilience.

One of the many parts of the transformation initiative for FFCU includes a data backup and recovery plan that ensures data is always available, always protected, and always in use, thanks to technology such as a cyber recovery vault.

Improved compliance, business growth, and enterprise-class business resiliency are among the many outcomes for this small, regional credit union. But what makes FFCU a great success story is that today, it offers cyber resiliency consulting to other federal credit unions, in addition to participating on technology advisory boards for cyber resiliency and digital transformation.

One more step: Practice

Another important step in ransomware defense that many organizations overlook is practicing their disaster recovery and response plans. Without running drills, simulations, and tabletop exercises, your team will have to work out the details in the middle of a crisis. That’s not the best time to figure out whom to call and where to find those phone numbers.

According to the GDPI survey, 67% of IT leaders are not confident they’ll be able to recover their business-critical data in the event of a destructive cyberattack. As an industry, we can do better. If you haven’t thought through the ransomware risks and implications yet, start that process now. With practice comes confidence. Be reassured: you don’t have to be beholden to brazen criminals. There are ways and means to protect yourself. Yes, at some point in time, you’ll be targeted (if you haven’t already). But you can choose how you respond and minimize the fallout. There are ways to protect your business and recover your data without submitting to the criminals’ demands and lining their pockets with your hard-earned money.

To learn more about achieving breakthrough transformation at the intersection of people and technology, visit Dell Technologies’ hub on MIT Technology Review here.

This content was produced by Dell Technologies. It was not written by MIT Technology Review’s editorial staff.

Concerns and anger over tech companies’ impact in the world is nothing new, of course. What’s changed is that workers are increasingly getting organized. Whether writing public letters, marching in protest, filing lawsuits, or unionizing, the labor force that makes the corporate tech world run is finding its voice, demanding a future in which companies do better and are held more responsible for their actions.

A week to remember

It began with a Facebook outage. For some six hours on October 4, 2021, services for its 3.5 billion users across the world were unreachable. The timing couldn’t have been worse for the company: just hours before, whistleblower Frances Haugen had dropped a series of damning revelations about Facebook’s willingness to put corporate goals above ethics and its users’ well-being. The stock price plunged. On the 5th, a Tuesday, Haugen would unflinchingly testify for three and a half hours before the United States Senate Commerce Committee on how “Facebook consistently chose to prioritize its profits” over public safety. 

If executives at Facebook and other tech companies hoped Haugen would be an outlier, Ifeoma Ozoma had other plans: a day after Haugen’s testimony, Ozoma and several colleagues launched the Tech Worker Handbook online. Ozoma was herself a whistleblower, having called out racial and gender discrimination at Pinterest, together with her coworker Aerica Shimizu Banks, in 2020. Since then, she has become something of an inspiration for whistleblowers in the tech world. “I’ve heard from tech workers asking for advice since I first went public,” she says. She responded to hundreds of people individually, but to her that solution was just “not scalable,” so she used what she’d learned from those experiences to build the website. It got 30,000 hits on the first day. 

The handbook guides potential whistleblowers on how to handle the media, explains legal rights, and teaches both online safety—to avoid corporate surveillance, for example—and offline tactics, like how to get through a doxxing campaign. “Preparedness is power,” says the front page. “Individuals should not have to rely on whisper networks for justice.” The site received an effusive response online and endorsements from activists, researchers, and other whistleblowers. 

Just a day after publishing her handbook, Ozoma notched another major victory for accountability: on October 7, California governor Gavin Newsom signed bill SB 331 into law. 

Also known as the Silenced No More Act, the bill protects workers who speak out about discrimination and harassment, even if they’ve signed a nondisclosure agreement, a common practice in the tech industry. The bill was written by state senator Connie M. Leyva and cosponsored by Ozoma, who drew from her whistleblowing and policy knowledge to help shape it. “Forty million people is a big fucking deal,” she says, referring to California’s population. “And if it would end there it would be a big fucking deal.” 

To support MIT Technology Review’s journalism, please consider becoming a subscriber.

It didn’t end there. As the law was making its way through the legislative system, a coalition of companies spearheaded by Ozoma pushed other tech firms to commit to extending its protection to all employees, not just those based in California. Expensify and Twilio agreed, but “it’s been a different story with Apple, Google, Facebook, Etsy, and a number of other companies,” Ozoma says.

Undeterred, the Transparency in Employment Agreements Coalition worked within the guidelines of the US Securities and Exchange Commission to file shareholder resolutions with seven technology companies, pushing them to extend the Silenced No More protections to all employees. Apple tried to get the proposal thrown out, but in late December the SEC ruled that the proposal does not “seek to micromanage the company,” as Apple claimed, meaning shareholders can now vote on it. If it’s passed at the March 4 annual meeting, the company will have to publish a public report on the use of concealment clauses in cases of discrimination or harassment. 

The Silenced No More Act went into effect on January 1, 2022. Even if every shareholder proposal effort fails, workers who live in California have been liberated from the restrictions that NDAs impose. The new law all but guarantees that new voices will step forward to bring their experiences to light. 

“All of the work that we’re doing and speaking up and organizing builds on what has come before and makes it possible for more to come in the future and to be successful,” Ozoma says.

Rooted in Techlash

To understand how advocacy and organizing within the tech industry work now, you have to go back to 2018, the year of the Techlash. Three important things happened that year. First a Cambridge Analytica whistleblower came forward with allegations of data misuse at Facebook. Then thousands of Google employees fought against Project Maven, an AI initiative created to enhance military drones. The year culminated in a massive, global Google walkout spurred by New York Times’ revelation of a $90 million exit payout to Android creator Andy Rubin following allegations of sexual misconduct. “The walkout, I think, cleared a space for everybody to scream in the streets,” says Claire Stapleton, one of the organizers. 

JS Tan and Nataliya Nedzhvetskaya help run the Collective Action in Tech archive, a project tracking the industry’s organizing efforts.

According to Collective Action in Tech, a project tracking the industry’s organizing efforts, every year since the walkout has seen more workers speaking out. The big tech companies’ image as friendly giants had been shattered. Part of the walkout’s legacy, Stapleton says, was “helping people see the gap between how companies present themselves and how they run a business, and what the capitalist machine is and does.”

In 2021, the sheer number of collective actions declined. But that’s because the nature of those actions shifted, say JS Tan and Nataliya Nedzhvetskaya, who help run the Collective Action in Tech archive.

“Compared to 2018, I think there’s a lot more realism about what organizing workers means and what comes with that,” says Nedzhvetskaya, a PhD candidate at the University of California, Berkeley. “One theory for why we’re seeing this base building is because people realize this is a hard thing to do individually.”

Last year, rather than penning open letters (which can be a fairly quick process), workers began pushing for unionization, a notoriously prolonged ordeal. But creating unions—even if they’re “solidarity unions,” which have fewer legal protections—is an investment in the future. Twelve tech-worker unions were formed in 2021, according to Collective Action in Tech’s analysis, more than in any previous year. Tan, who originally conceived the archive, says most of those unions are at smaller outlets where there are fewer barriers to organization. But workers from larger firms are getting in on the action too. 

“If the goal is to hold these big tech companies accountable,” says Tan, himself a former tech worker who helped organize within Microsoft, “it’s not just one of these groups of workers who’s going to be able to do it. It’s the combined strength of them.”

The fight against “digital slavery”

Nader Awaad knows where to find Uber drivers with time to spare. He approaches them while they idle in the parking lots outside London’s bustling airports, waiting for customers. Awaad hands them a leaflet and talks to them about joining a union, patiently hearing them make the same complaints he’s heard echoed across the industry. 

Gig drivers aren’t the white-­collar software developers you might picture when you think of a tech worker, but they make up a huge and growing group of tech employees. Over the last year, they have become increasingly vocal about several basic demands: for better pay, increased safety, a way to seek recourse if they are unfairly kicked off a company’s app. In the UK and South Africa, drivers have brought Uber to court. In the US, DoorDash drivers went on an unprecedented, countrywide strike over plunging pay. In Hong Kong and mainland China, food delivery workers organized strikes for better pay and safety. In Croatia, Uber drivers held a press conference and a strike, saying their payments were late. “We feel like digital slaves,” one union member said. 

WIKTOR SZYMANOWICZ/NURPHOTO VIA AP

Awaad began driving for Uber in 2019 after being laid off from his previous job as a senior manager. He immediately felt the industry’s problems. “It reminded me of reading Charles Dickens,” he says. “The level of exploitation. The level of deprivation. I said, ‘I can’t believe it.’” Just as quickly, he realized he was not alone. Another driver he met at Heathrow sympathized. He looked around for a union to join, and by April 2019 he was a member of United Private Hire Drivers, a branch of the Independent Workers Union of Great Britain. He is now the elected chair. 

His local membership of 900 or so drivers echoes those global problems, and he’s helped organize pickets and strikes, but he says the companies are refusing to engage in open dialogue. Awaad says drivers have to stay on the road for 12 or 14 hours a day to earn enough to get by. 

In a landmark case last February, the UK’s Supreme Court ruled that drivers are entitled to holidays, pensions, and a minimum wage. Several unions say Uber has avoided those new obligations, but the European Commission has also taken notice of the problem. It issued a directive in December to “improve the working conditions in platform work,” meaning new rules will follow. 

COURTESY PHOTO

Then there’s the problem of algorithmic discrimination. Companies use algorithms to verify that drivers are who they say they are, but face-recognition technology is notoriously worse at recognizing nonwhite faces than white ones. In London, the vast majority of drivers are people of color, and some are getting removed from the platforms because of that gap. 

Termination without a chance for appeal was a major motive for a strike Awaad helped organize in October. About 100 drivers rallied in the brisk London air, holding a large black banner with “End unfair terminations, stop ruining lives” written in white. In the background, protesters held signs with photos of drivers. “Reinstate Debora,” one of them said. “Reinstate Amadou,” said another. 

During that rally, United Private Hire Drivers announced a discrimination complaint it had filed on the basis of the face-recognition errors. “We expect the court to come heavy on Uber because it happens in other countries, not only in our country,” Awaad says.

 “At first I don’t think I understood how big the moment was going to be,” Field says. By the afternoon, big-name celebrities were voicing their support.

The drivers who do get work face other dangers. Covid exposure is an ongoing concern. So is assault—Awaad has spoken with  drivers who have been attacked and robbed of their cars. He plans to organize a protest in front of the UK parliament to demand safety measures, and has been reaching out to other unions representing drivers, hoping to form a coalition and get the companies to act. 

“We have two drivers who were killed in Nigeria. We have a driver who was killed on the 17th of February in London. We have, on a daily basis, attacks against the drivers,” Awaad says. “It’s not something that has to do with London only. It’s a global issue.”

Busting union busters

In September, workers at Imperfect Foods who had voted to unionize found that their employer was prepared to play the role of union buster. The same thing happened in November at HelloFresh, another grocery delivery service, whose workers in Aurora, Colorado, reported bullying and intimidation from management. When workers at an Amazon warehouse in Alabama held a vote in April on whether to unionize, the company interfered so extensively that the US National Labor Relations Board ordered a do-over. (In a separate settlement, the agency said Amazon must allow its workers to freely organize unions.) 

Such tactics are spreading, according to Yonatan Miller, a volunteer with the Berlin chapter of the Tech Workers Coalition. “Germany has a strong tradition of social compromise and social partnership, where companies are not as adversarial or hostile,” Miller says. “This is something that you’re kind of seeing imported from the US—this kind of US-style union-busting industry.” 

ULI KAUFMANN

The Tech Workers Coalition is a grassroots, volunteer-led organization with 21 chapters globally. Miller got involved in 2019 and still remembers the first meeting, in Berlin’s Kreuzberg neighborhood, with about 40 tech workers in attendance. “Most of us were, as they say in Germany, newcomers. And some of us were from Arabic or Muslim background,” he says. But most were from Latin America, Eastern Europe, or elsewhere in Europe.

The idea behind the coalition is to help find a global answer to a global problem, and in the Berlin chapter’s two years of operation, it has achieved plenty of tangible results. It helped organizers at the grocery app Gorillas, Germany’s first unicorn company, which fought bitterly against a workers’ council, a union-like organization within a company that negotiates rights for workers. It also helped raise funds for an Amazon warehouse worker in Poland who was fired in what the coalition says was retaliation for her union activity. When the HelloFresh workers were trying to unionize, the coalition chapter in Berlin organized a protest in front of the company’s headquarters in solidarity. Any time there’s need or want, the coalition comes in to provide training, advice, or support, much of it “happening more discreetly behind the scenes,” Miller says. 

In his eyes, these efforts are bringing the tech industry closer to other industries’ standards. His labor organizing is inspired as much by the activity of teachers and health workers as it is by the Google walkout. The inability to mingle with these other workers is one reason the pandemic has been so frustrating—it cut off access to the bars and gatherings where complaints turn into ideas and, eventually, actions at a moment when the industry had just begun to accept the need for labor organizing. “We won the moral argument,” Miller says, “but we haven’t been able to flex it.”

Tech, with integrity

The dust from Frances Haugen’s testimony last October hadn’t yet settled when two former Facebook workers made an announcement. Sahar Massachi and Jeff Allen were launching the Integrity Institute, a nonprofit intended to publish independent research and help set standards for integrity professionals, who work to prevent social platforms from causing harm. Both Massachi and Allen had been ruminating on the idea for a while. They’d worked to clean up platforms as part of Facebook’s integrity team; some of Allen’s research was among the documents Haugen leaked. Now they wanted to answer big questions: What does integrity work look like as a discipline? What does it mean to responsibly build an internet platform? 

Sahar Massachi and Jeff Allen launched a nonprofit to publish independent research and help set standards for integrity professionals.

“When you’re doing that work, it just very quickly becomes clear that this is going to be a very long-term problem,” says Allen. It’s also a problem that goes beyond Facebook. Every platform struggles with how to handle spam, foreign influence operations, and networked disinformation. Allen and Massachi want the Integrity Institute to be the go-to place for advice and original research on these issues, which they plan to publish in an open-source format. 

“Hopefully we’ll get to a place where we can say being a tech worker is an ethical practice,” Allen says.

Massachi and Allen are not whistleblowers. They’re careful to stay well within the bounds of their NDAs and avoid getting into detail about their time at Facebook. But they represent a larger trend of former big tech employees using their expertise to bring knowledge about platform functions into the public light. 

This includes algorithm ethicists, who had their own revolution in 2021. Meredith Whittaker, an AI researcher and former Google employee who helped organize the 2018 walkout, is now advising the US Federal Trade Commission. Timnit Gebru, who was fired in December 2020 from her position co-leading Google’s Ethical AI team, announced the creation of the Distributed AI Research Institute a year later. 

The firing of Gebru and then the team’s founder, Margaret Mitchell, had sent shock waves through the AI and tech communities. Google employees penned a letter of protest to CEO Sundar Pichai, other engineers resigned, and a campaign titled #MakeAIEthical sought to disrupt Google’s influence over the field. “Right now, it’s obviously very difficult to imagine how anybody can do any real research within these corporations,” Gebru told MIT Technology Review in December 2020. “But if you had labor protection, if you have whistleblower protection, if you have some more oversight, it might be easier for people to be protected while they’re doing this kind of work.”

“Strength in community”

Terra Field worried it would rain on the day of the walkout. The weather was cloudy, and she wasn’t sure how many people would show. “None of us were particularly amazing at project management,” she says. At Netflix, there’s an ongoing joke that every meeting starts five minutes late: it’s called the Netflix Five. That was true of the walkout, too. Field watched as, at five minutes past the hour, the parking lot of Netflix’s office in Los Gatos, California, filled with trans workers and their allies, gathering to rally against the company’s reaction to criticism of Dave Chappelle’s latest special, The Closer.

Field hadn’t expected her Twitter thread about the comedian’s anti-trans jokes to lead to the work stoppage, but it was “incredibly gratifying” when over 100 people came to the rally. “At first I don’t think I understood how big the moment was going to be,” she says. By the afternoon, big-name celebrities—Dan Levy, Elliot Page, and Jonathan Van Ness among them—were voicing their support on social media. 

COURTESY PHOTO

Field began working at Netflix in 2019; it was her first job after she transitioned. She talked to the trans employee resource group during the interview process, and after she was hired, it wasn’t long before she joined herself. The group, she said, became a lifeline for many during early pandemic isolation, and especially for those who transitioned during that time and needed community. But it also served a larger function at Netflix. Members met with other teams to provide training. “It was a lot about helping people understand the trans experience—trying to build empathy, understanding,” Field says. 

One of the key relationships the group built was with the content teams, who turned to its members as a resource and a sounding board. “It meant that there weren’t things that accidentally might hurt the trans community,” Field says. It was an informal process, but a helpful one. The group helped consult on a much-praised episode of The Baby-Sitters Club, for example, which thoughtfully featured a trans character. It also helped with questions on how to dub trans voices in different countries and talked about “how to make the content more authentic.” 

But the group was blindsided by Chappelle’s special. Its members found out at the same time as everyone else—through a push notification. “We kind of felt betrayed on some level,” Field says. She thinks the reaction would have been different had trans workers known it was coming and had the opportunity to give feedback. “I’m from New Jersey, so the way I deal with things is complaining about them loudly,” she says. 

That’s when Field sent her series of mega-viral tweets. Following the thread, she was suspended for attending a high-level meeting the company said she was not supposed to be at. After the walkout, she resigned, citing the firing of B. Pagels-Minor, another trans employee who helped organize it. Pagels-Minor denied having leaked sensitive materials to the press, as Netflix claimed. 

“At first I don’t think I understood how big the moment was going to be,” Field says. By the afternoon, big-name celebrities were voicing their support.

According to data from Collective Action in Tech, identity-based discrimination was a driving factor for much of the current wave of organizing at US tech companies. Workers have demanded that companies remove anti-Asian content from their platforms and penned letters calling on corporate leaders to support Palestinians. Over 2,000 Apple employees petitioned against the recent hiring of an executive who they said was “misogynistic” and held “harmful views” on women and people of color. The executive left the company as a result. During last year’s pride month, Google workers circulated a petition asking the company to take steps to ensure the use of chosen names instead of birth names for trans employees.

It makes sense to Field that Silicon Valley companies don’t see more protests about wages from their white-collar employees—those workers get stock options, good salaries, and free lunch. But such perks do little to address structural discrimination. 

Field says she can’t imagine this type of action having happened five or 10 years ago, but now the gates have been flung open. “My hope is that this momentum continues and expands,” she says, “as people realize there’s strength in community.”

Jane Lytvynenko is a senior research fellow on the Tech and Social Change Project at the Shorenstein Center at the Harvard Kennedy School.

Chemists at Pfizer’s research facility in Connecticut dusted off some ideas the company had developed during the SARS outbreak in 2003. Even back then, an obvious line of attack had been to block a well-understood component of the virus life cycle involving a key protease, a protein that orchestrates how the virus copies itself. Find a chemical that is able to stick tightly enough to that protein, and it would stop the virus from replicating in the body, lessening the chances that a patient would become seriously ill.

Right away, researchers got a lucky break. When Pfizer checked, it found that none of the thousands of proteins in the human body shared the same bit of molecular structure they planned to interfere with in SARS-CoV-2. That meant they could hit the virus hard and not expect any major side effects. Nature had provided the scientists with a big bull’s-eye. “This is the most solid biological target I have ever worked on,” says Pfizer chemist Dafydd Owen.

Pfizer sped the drug forward by testing hundreds of chemicals in parallel and then making big batches of the most promising one. Even as the first covid vaccines were authorized in the US in December 2020, animal studies were underway on the drug that would later be named Paxlovid. Human trials began in March 2021.

By the fall of 2021, Pfizer was ready to declare success. A monitoring board decided to stop the human study because covid-19 patients on Paxlovid weren’t dying—but those given the dummy drug were. “It was an incredible moment,” says Charlotte Allerton, Pfizer’s head of medicine design. Even though it trailed the vaccine development by nearly a year, Allerton believes Paxlovid still set a record—the fastest any drug company has ever moved from a synthesizing a new chemical to proving that it safely treats a disease. 

A test in unvaccinated volunteers done by Pfizer had shown that the new pill cut the chance of a serious case of covid by 89%. And the results seemed to come at a perfect time. Infections and deaths were about to reach new heights. The new, fast-spreading omicron variant has infected millions of people each day just in the US. President Joe Biden, whose administration authorized the pill’s sale on December 22, 2021, touted it as “a game changer.” 

So far, the world has looked to vaccines for prevention and, in rich countries, to expensive IV infusions of drugs called antibodies that block the virus. With pills in blister packs that you can pick up at midnight at the pharmacy with a prescription, there will be what the doctor and social media pundit Eric Topol calls “a whole new approach to tackling the virus.”

“You get a prescription, you go to the CVS, and that’s it.”

Crucially, the protease is, in the jargon of biologists, a “highly conserved” molecule. That means that even as the virus evolves, this part rarely changes. So while the coronavirus has been mutating quickly to evade vaccines, so far it looks as though Paxlovid will work just as well against any variant—whether it’s omicron or whatever comes next. 

In fact, laboratory tests run by Pfizer suggest Paxlovid will work against all coronaviruses, maybe even one still lurking in a bat cave somewhere. If so, it means the company has hit on a potential defense against the next outbreak, too. “It has the potential to be a pan-coronavirus agent and stockpiled against future pandemics,” says Owen, the Pfizer chemist. “But it’s here for this pandemic, because we did it super fast.” 

To support MIT Technology Review’s journalism, please consider becoming a subscriber.

Pfizer’s drug isn’t the only antiviral to show promise. In late 2020, a drug called remdesivir was the first chemical approved in the US to treat covid-19. But remdesivir has to be given through an IV drip, five days in a row. That has limited its impact. In contrast, Pfizer’s chemists tweaked their antiviral chemical so you could swallow it. 

“I feel that Paxlovid is the big step we were working for this pandemic,” says Kris White, a researcher at the Icahn School of Medicine in New York, who was recruited by Pfizer to give the drugs to mice in 2020. “I believe it is going to be the treatment for covid.” 

As he says, “You get a prescription, you go to the CVS, and that’s it.” 

Cautious optimism

Despite the early excitement, Pfizer’s pill still remains in short supply. 

Desperate to end the pandemic, the Biden administration immediately spent $5.3 billion to pre-purchase 10 million courses of Paxlovid in December and doubled the amount a few weeks later. But those 20 million courses won’t all become available until midyear, mostly too late to deal with the current omicron surge. 

And some medical researchers suspect Pfizer’s spectacular test results could be too rosy. The human trial that led to its authorization was relatively small, involving about 2,000 people, which means the true effectiveness of the drug could turn out to be less stellar in the real world. “We should not jump to conclusions about its miraculous efficacy,” says Thomas Agoritsas, a doctor specializing in medical evidence at the Geneva University Hospitals. 

NICO ORTEGA

Another drawback is that Paxlovid should be given within five days of the start of symptoms. Pfizer’s own internal models identify that as a challenge. An August 2021 study in the Annals of Emergency Medicine found that, on average, people have symptoms for five or six days before they turn up at a hospital. By that time, those with serious cases are gasping for air and face deadly lung problems due not to the virus, but to their body’s immune reaction against it. At that point, the pill can’t help. 

That raises questions about whether Paxlovid will actually ease the pandemic. Even when patients are not that sick, there’s often a time lag while their infection is confirmed. For this reason, Pfizer has floated the idea of offering the drug to people while they wait for test results.

“The name of the game is speed,” says Myoung Cha, president of home-based care at Carbon Health, which operates walk-in medical clinics in the US. “Even if we had oral drugs available today, the testing debacle would prevent many people from getting treated.” 

Pfizer is also running a study to see if the pills help people who’ve only been exposed to covid-19, as a sort of prophylactic treatment.

“It’s a tight window: two days to get tested and another two to get the drug.” 

For now, there isn’t enough Paxlovid to go around, so the drug is being rationed—and so far in a chaotic fashion. The US Food and Drug Administration authorized the pills for anyone with a confirmed covid-19 infection and one risk factor for developing serious illness. But which risk factors qualify—and which patients should get the drug—is still up for debate. 

Bob Wachter, head of medicine at the University of California, San Francisco, announced on Twitter that his hospital would be reserving the pills for people with compromised immune systems, like kidney transplant or cancer patients. The state of New York suggested that it might prioritize Black and Hispanic residents, reasoning that they are at higher risk because of health inequities. 

The most significant risk factor for serious covid-19 is being unvaccinated—and it was unvaccinated people whom Pfizer studied in its human trial. If avoiding the shots puts you near the head of the line for the pills, people could take that as reason to stay unvaccinated. However, David Boulware, a doctor who studies covid-19 treatments at the University of Minnesota, suspects that people who refuse the vaccine might not be seeking treatment in time to get Paxlovid. He says patients in his hospital’s ICU are mainly vaccinated people with abnormal immune systems, or unvaccinated people who turn up short of breath and are already in considerable distress. Some have already tried “random” home remedies or disputed treatments like the antiparasite drug ivermectin. 

“It’s a tight window: two days to get tested and another two to get the drug,” he says. “If you are sitting at home and think covid is a hoax, will you get tested quick enough? Because by the time you’re in the hospital, your disease is being driven by the body’s inflammation response and by then the antivirals don’t have a big role.”

In a statement, the World Health Organization said it believes “prevention is better than cure” and that “these drugs will not be alternatives to vaccines.” The organization, based in Geneva, has yet to make a formal recommendation in favor of Paxlovid and says it wants to track whether side effects emerge.

“It’s going to be very hard to use Paxlovid on a wide scale, because people are going to have to be tested and treated very early,” says Robert Shafer, a professor of medicine at Stanford University. “It’s just not going to have the same impact that vaccines will, and it will be a very expensive solution in comparison.” 

A different strategy

Maybe so. But the pills are still an important addition to the anti-covid arsenal.

Early in the pandemic, international organizations plowed billions into vaccine programs. They also gave priority to “repurposing” existing drugs, essentially searching pharmacy shelves for anything that might help. But designing a new, customized chemical drug didn’t get the same kind of public support. “The world seemed to give up on new antivirals before they even started,” Annette von Delft, a researcher at the University of Oxford, wrote in Nature last year. 

Von Delft is part of an organization called Covid Moonshot that says it struggled to find funding for new antiviral pills. That’s despite some big successes with other antivirals, like the pills that keep HIV in check and, more recently, those that conquered hepatitis C. The group says one reason is that health authorities believed designing a new chemical from the ground up would take too long.

It’s true that such an effort involves unavoidable rounds of trial and error. “You can’t give a computer an enzyme and say, ‘Design me a drug for this.’ It might give you 100 ideas, but then you have to synthesize those,” says Michael Lin, a researcher at Stanford University. Synthesizing a single drug can take several weeks, and then you still have to learn its key properties, like whether it’s absorbed in the gut or broken down in the liver. All that is done through real-life tests on animals. 

What’s more, some large drug companies have shifted away from antiviral research in recent years. Despite the successes with HIV and hepatitis C, the list of viruses affecting rich countries—viruses for which there’s no vaccine and where a pill could make money—hasn’t been very long. Academics like Icahn’s White, who is a specialist in influenza drugs, saw their career prospects dimming. “People didn’t think there were any more profitable viruses to treat,” says White. “There was a period there where it was hard to stay in business.”

But, it turns out, the chemists knew a few tricks that proved invaluable against the covid virus.

SARS-CoV-2 causes illness by injecting a cell with genetic material that gets the cell to copy the proteins needed to manufacture even more virus copies. As it turns out, a number of those viral proteins are generated as one long piece—think of a chain of connected sausage links. The job of the protease chemists were targeting is to cut this big “polyprotein” into working parts, something it does using a special molecular notch. 

The researchers knew if they could stuff that opening with a chemical that sticks to it very strongly—so strongly it can’t be detached—the protease wouldn’t do its job and the virus wouldn’t multiply. “To create a protease inhibitor, it’s like milling a key to fit a lock,” says Lin. “You want a drug that fits in that pocket perfectly and makes it unavailable.”

By mid-2020, chemists including Lin were tossing out proposals for chemical shapes that might work. But making and testing chemicals in a hurry is where the limitless R&D funds of big companies come into play. Pfizer was able to synthesize 800 molecules in all, according to the company. 

After identifying the most promising ones, in September 2020, the company moved quickly. At that point, a company would normally invest in small production amounts for testing. But Owen wanted enough drug on hand to start a human study right away if it worked in animals. He took the gamble of speeding up production. 

By December 2020, some of the first supplies of the new compound had reached White in New York. All eyes were on the vaccines from Moderna and Pfizer, which were approved that month. But in White’s lab, he was clearing his schedule so that Pfizer’s antiviral drugs could be given to mice infected with SARS-CoV-2. “I was extremely busy when Pfizer emailed, but we set up a Zoom and I moved them to the head of the line,” he says. 

The first compound from Pfizer he tried was a bust. The second, Paxlovid, was an obvious hit, reducing the amount of virus in the mice’s bodies by a factor of one thousand or more. Within a year, the drug had gained authorization from the FDA.

Cost effective

In purchasing 10 million courses of Paxlovid for $5.3 billion, the US established a price of around $530 per course of pills—six tablets a day for five days. Italy, Germany, and Belgium also placed orders. According to Pfizer CEO Albert Bourla, the price of the vaccine (around $30 for each dose) helped determine what the new drug should cost. 

For Pfizer, easy-to-take covid-19 pills could become another blockbuster. “It’s a license to make money. As much of it as they can make, they can sell to governments,” says Boulware. “Demand will outstrip supply, and that is going to be the case for the foreseeable future.”

But even at around $500 per person, Paxlovid could be a bargain. If Pfizer’s trial numbers stand up, doctors who give the drug to the patients at greatest risk could save about one person for every 100 they treat. That’s $50,000 for a life. Medical economists say the drugs even have the chance of being “cost negative”—that is, they’ll save money if they keep enough people out of the hospital, since each hospitalization costs thousands of dollars. 

One area where antiviral pills have an edge is as an insurance policy against new variants—or even different coronaviruses that are yet to be discovered. Covid-19 has surprised scientists again and again by mutating in ways that allow it to spread faster or even evade immunity. Of the antibody drugs authorized to treat covid-19, several, such as the one sold by biotech company Regeneron, no longer work against omicron.

NICO ORTEGA

Resistance like that occurs because the virus continually changes its “spike”—the molecule it uses to get into cells, and the one targeted by vaccines and antibodies. Being able to shape-shift the spike gene, which is the most exposed part of the virus, is probably an evolutionary survival strategy—one that lets coronaviruses adapt to new species and dodge immune reactions. But researchers don’t think the virus can so easily evolve ways of dodging Paxlovid. That’s because the protease is very finely tuned for its job, so much so that even distantly related viruses have proteases that look very similar. 

Could a Paxlovid-resistant form of covid-19 appear? It could—the protease could conceivably evolve enough to dodge the drug. But such a variant might be less good at copying itself and probably wouldn’t spread very far. “I don’t think that resistance is a big concern,” says Shafer, who maintains a database of drug-resistant types of HIV at Stanford University. “Changes to the protease are bad for the virus.” A treatment that lasts just five days also doesn’t give much time for the virus to evolve resistance, he says.

The less changeable nature of the protease gene—even among cousin germs—is also why Pfizer’s drug might prove useful against viruses we haven’t encountered yet. Laboratory tests run by the company show that in addition to blocking the growth of SARS-CoV-2 in cells, it also inhibits half a dozen other coronaviruses. These include MERS, a dangerous germ spread by camels that kills a third of the people who get it; the original SARS virus from 2003; and a handful of coronaviruses that cause only colds. 

And although Paxlovid is the most promising antiviral out there for covid-19 right now, more than a dozen new antivirals are now in development; the next generation could be even better. That’s what happened with HIV. There are now so many effective HIV drugs on the market that the original protease inhibitors have been relegated to second-line treatment. 

New pills for other viral diseases could be on the way too. In June 2021, the US finally turned its attention back to antivirals in a big way, announcing it would spend $3 billion on a major search for next-generation drugs. About half that money will pay to establish eight to 10 new antiviral research centers that will each work on covid-19 and another germ of their choice, like Ebola or the common cold. 

“The espoused goal is to have something that’s off the shelf the next time one of these major public health threats emerges—actually, we want to have many things,” says Matthew Frieman, a coronavirus specialist at the University of Maryland School of Medicine, who is among those applying for the funds. According to Frieman, the idea is that coronaviruses could one day be treated with a combination of antiviral drugs, similar to the “cocktails” used to control HIV. “It’s the same idea: the more drugs in combination, the better,” he says. “And you protect against mutant viruses, because it’s harder to escape from two drugs than one. I think we need a suite of antivirals that target this virus.”

It may even be possible to find drugs that work against nearly any virus, even ones as different as Ebola and influenza. Frieman says he’s found some compounds that may do that by acting on the human body, rather than on parts of the virus. “We’re hoping there are a whole new spectrum of ways to target viruses,” he says. “We just need to find them. In the past, we had no funding because no one cared. I think we have only scratched the surface.”

Antonio Regalado is MIT Technology Review’s senior editor for biomedicine.

The findings: The 36 volunteers, all aged 18 to 30, were exposed to a low dose of the original SARS-CoV-2 virus in the nose, the equivalent of the amount found in just a single drop of nasal fluid. Half the participants developed covid symptoms; they became infectious within just two days, with levels of infectious virus peaking at five days. It has previously been estimated that the time from exposure to first symptoms was about five days. Participants in the study remained infectious for an average of nine days and still had detectable levels of virus in their nose 12 days after initial exposure.

Almost all the volunteers lost their sense of smell and experienced cold-like symptoms such as a runny nose and sore throat. None reported serious symptoms.  Some of the patients were also given the antiviral drug remdesivir before they were infected, but the trial didn’t pick up any noticeable difference in the severity of the symptoms.

What it all means: The findings come with the caveat that they’ve been derived from a small pool of volunteers, and they were published in a preprint paper that has not yet been peer-reviewed. However, they provide useful insights nonetheless. The fact that people become infectious so quickly and stay infectious for so long suggests that recommended isolation periods should be kept at around 10 days. Although the virus was detected in the throat first, it was eventually present at much greater levels in the nose, highlighting the need to wear face masks properly so they cover the nose.

Get tested: The research also supports the regular, widespread use of lateral flow testing. Modeling using the study data found that regular rapid tests can diagnose infection before 70% to 80% of infectious virus had been generated, meaning that if people get tested regularly and isolate when positive, it could significantly cut community transmission. The fact that none of the participants fell severely ill also suggests this trial method could be used to test future variants or drugs. 

Michael Jacobs, a consultant in infectious diseases at the Royal Free London hospital, where the trial was carried out, said in a statement: “The trial has already provided some fascinating new insights into SARS-CoV-2 infection, but perhaps its greatest contribution is to open up a new way to study the infection and the immune responses to it in great detail and help test new vaccines and treatments.”

This phenomenon is not unique to AI. In today’s fast-moving digital economy, it’s not uncommon for performance and results to lag behind expectations. Despite the enormous potential of modern technology to drastically improve the customer experience, business innovation and transformation can remain elusive.

According to Gartner, 89% of companies now compete primarily on the experiences they deliver. As marketers and other teams turn to these systems to automate decision-making, personalize brand experiences, gain deeper insights about their customers, and boost results, there’s often a disconnect between the technology’s potential and what it delivers.

When it comes to AI, frequently, organizations fail to realize the full benefits of their AI investments, and this has real business repercussions. So how do organization ensure that their investments deliver on their promise for fueling innovation, transformation, and even disruption? To find success, it requires the right approach to operationalizing the technology, and investing in AI capabilities that can work together throughout the entire workflow to connect various thoughts and processes together.

Getting real about AI. Realizing the value of AI starts with a recognition that vendor claims and remarkable features will only go so far. Without an overarching strategy, and a clear focus on how to operationalize the technology, even the best AI solutions wind up underperforming and disappointing.

 There’s no simple or seamless way to implement AI within an organization. Even with powerful customer modelling, scoring or segmentation tools, marketers can still wind up missing key opportunities. Without ways to act on the data, the dream of AI quickly fades. In other words, you may know that a certain customer likes hats, and another customer enjoys wearing scarfs but how do you move these people to an actual purchase, or deliver the right content for where they’re at in the buying lifecycle?

The winning approach is to start small and focused when it comes to implementing AI technology. Be mindful about what types of data models you can build with AI, and how they can be used to deliver compelling customer experiences, and business outcomes. Collecting and analyzing actionable customer data is only a starting point. There’s also a need to develop content that matches personas and market segments and deliver this content in a personal and contextually relevant way. Lacking this holistic view and AI framework, organizations simply dial up speed—and inefficiency. In fact, AI may result in more noise and subpar experiences for customers, and unrealized results for an enterprise.

Moving from transaction to transformation. A successful AI framework transforms data and insights into business language and actions. It’s not enough for the marketing team to know what a customer likes, for example, it’s essential to understand how, when and where an individual engages with a business. Only then, can a brand construct and deliver a rich customer experience that matches the way their customers think about and approach a brand. This includes an optimal mix of digital and physical assets, and the ability to deliver dynamic web pages, emails, and other campaigns that customers find useful and appealing. When a marketer understands intent and how a person travels along the customer journey, it becomes possible to deliver the most compelling customer experience.

With this framework in place, marketers can read the right signals and ensure that content delivery is tuned to a person’s specific behavior and preferences. It’s possible to send emails, serve up ads and mail brochures that reach consumers when they are receptive and ready to engage. Whether the customer is into hats, scarves or electric guitars, the odds of successful marketing increase dramatically.

Putting AI to work. Only when an organization has mapped AI workflows and business processes—and understands how to reach their customers effectively—it’s possible to get the most out of AI solutions. These solutions can address the full spectrum of AI, including reading signals, and collecting, storing, and managing customer data; assembling and managing content libraries; and marketing to customers in highly personalized and contextualized ways.

A good way to think of things is to imagine that a person hops in a car with the intent of driving across the United States. If the journey is from Los Angeles to New York, for example, it’s tempting to think the motorist will take the most direct route available. But what happens if the person loves nature and wants to visit the Grand Canyon or Yellowstone National Park along the way? This requires a change in routing. Similarly, an organization must have the tools to understand how and where a person is traveling in the product lifecycle, what ticks the person’s boxes along the way, and what helps them arrive at a desired destination with a minimum of friction and frustration.

AI can do this—and it can serve up promotions and incentives that really work. Yet, to build the right customer experiences and the right journey, marketers must move beyond AI solutions that deliver a basic customer score or snapshot, and instead obtain a motion picture-like view of a customer’s thinking, behavior, and actions. To that end, building out one AI capability or buying one point technology to address a single aspect of customer experience isn’t enough. It’s about being able to connect a set of AI capabilities, which are orchestrated throughout the entire workflow to connect various thoughts and processes together.

Only then is it possible to deliver an optimal marketing experience.

Delivering on the promise of AI. To be sure, with the right strategy, processes, and AI solutions, it’s possible to take marketing to a more successful level and deliver winning customer experience. When marketers truly understand what a customer desires and how they think about a product and their customer journey, it’s possible to tap into the full power of AI.

What’s more, this approach has repercussions that extend far beyond attracting and retaining new customers. When organizations get the formula right, marketers can engage with their best customers in a more holistic and natural way. In the end, everyone wins. The consumer is greeted with a compelling customer experience with relevant messages that display products and services they are interested in at every step of their journey and the business boosts brand value and loyalty.

At that point, AI finally delivers on its promise.

If you’d like to learn more about how AI can help your company deliver personalized content at scale, visit here.   

This content was produced by Adobe. It was not written by MIT Technology Review’s editorial staff.

Herrera is the newly minted leader of the National Security Agency’s Research Directorate. The directorate, like the rest of the NSA, has a dual mission: secure American systems and spy on the rest of the world. The budget is classified, a secret among secrets, but the NSA is one of the world’s largest spy agencies by any measure and Herrera’s directorate is the entire US intelligence community’s biggest in-house research and development arm. The directorate must come up with solutions to problems that are not yet real, in a world that doesn’t yet exist. 

In his first interview since getting the job, Herrera lays out the tech—and threats—his group will now be focusing on. His priorities show how much the NSA’s targets are changing, balancing its work surveilling terror groups with an appreciation of how rapidly the geopolitical landscape has shifted in recent years. And he explains why the rise of new technologies, in terms of both threat and opportunity, are at the heart of what his group must contend with.

Herrera takes the helm as the agency faces new challenges. The bipolar world of the Cold War belongs to the history books. The United States’ quick turn as a lone superpower is over. The new world is a messier one, defined by an emerging era of great power competition among nations like the United States, China, and Russia. Meanwhile, the NSA is still recovering from a massive set of leaks published nine years ago about global and domestic surveillance programs that set off a firestorm of criticism and calls for reform and changed the average American’s perception of the NSA. The companies that worked with them recoiled in embarrassment and anger. And it also changed the way the NSA operates.

“We’re at a point now where we need to start focusing more on larger adversaries, more sophisticated adversaries, adversaries that don’t necessarily utilize commercial services,” says Herrera. “These are adversaries that have their own indigenous services and that create their own technology. So as a research directorate, we need to respond. We need to provide the technologies that allow us to interrogate the huge amounts of information brought to us and to help monitor the kinds of systems that are now emerging as a result of great power competition.” 

The rate of technological change is accelerating and becoming less predictable. 

“Any time there’s that kind of a shift, it’s complex,” says Herrera. “Each generation of technology presents its new challenges.” 

For example, the directorate has devoted significant resources toward mastering quantum computing, technology that has the potential to break the encryption used to protect sensitive data in the digital world of today and tomorrow. Powerful countries, companies, and universities are pouring money into the task of building a quantum computer powerful enough to perform exponentially faster than the computers of today.

“Great power competition drives the agenda,” says Herrera. “It changes the kind of technology and access we need. Technologies like quantum and 5G are part of that.”

The directorate has been at the forefront of quantum computing research since 1995, immediately following the advent of Shor’s algorithm, which showed how quantum computers can factor numbers exponentially faster than normal computers—exactly the kind of work needed to break encryption.

The directorate’s fingerprints now show up in the form of fundamental research advancing the field and even inside the most advanced computers built at giant tech firms. The highly publicized race to build the world’s best quantum computer is proof of this: both Google and IBM use the same basic building block in their machines to create quantum behavior, known as transmon qubits, which was invented under the directorate’s sponsorship. Historically, the NSA has been the single largest funder of academic quantum computing research, says Herrera. 

Herrera is hesitant to discuss specifics about what his directorate is zeroing in on, but when asked about the challenges of spying in a world of rapid technical advancement, he agrees and points to the emergence of 5G around the world. 5G brings its own new challenges for collecting intelligence, Herrera explains. Monitoring 5G successfully requires a deep understanding of what makes it fundamentally different from its predecessors: higher speed, lower range, more distribution nodes, different data protocols.

Understanding what will happen in the world tomorrow requires a mastery of the elements that will define it.

Future history

The NSA’s Research Directorate is descended from the Black Chamber, the first group of civilian codebreakers in the United States who were tasked with spying on cutting-edge technology, like the telegraph. Existing only from 1919 to 1929, the group decoded over 10,000 messages from a dozen nations, according to James Bamford’s 2001 book Body of Secrets: Anatomy of the Ultra-Secret National Security Agency. In addition to groundbreaking cryptanalytic work, the group succeeded by securing surveillance help from American cable companies like Western Union that could supply the newly minted US spies with sensitive communications to examine. 

The Black Chamber was shut down amid scandal when US Secretary of State Henry Stimson found out the group was spying on American allies as well as foes. The incident foreshadowed the 1975 Church Committee, which investigated surveillance abuses by American intelligence agencies, and the 2013 Snowden leaks, which exposed vast electronic surveillance capabilities that triggered a global reckoning. 

Just eight months after the Black Chamber was shuttered, the US, faced with the prospect of crippled spying capabilities in the increasingly unstable world of the 1930s, reformed the effort under the Army’s Signals Intelligence Service. One of just three people working with the Black Chamber’s old records, one of the founders of the SIS, which Bamford reports was kept a secret from the State Department, was the mathematician Solomon Kullback. 

Kullback was instrumental in breaking both Japanese and German codes before and during World War II, and he later directed the research and development arm of the newly formed National Security Agency. Within a year, that evolved into the directorate as we know it today: a distinct space for research that is not disrupted by the daily work of the agency.

“It’s important to have a research organization, even in a mission-driven organization, to be thinking beyond a crisis,” says Herrera, though he adds that the directorate does dedicate some of its work to the “crisis of the day.” It runs a program called “scientists on call,” which allows NSA mission analysts facing technical challenges while interrogating information to ask for help via email, giving them access to hundreds of scientists.

Forward looking

But the lion’s share of the directorate’s work is envisioning the technologies that are generations ahead of what we have today. It operates almost like a small, elite technical college, organized around five academic departments—math, physics, cyber, computer science, and electrical engineering—each staffed with 100 to 200 people.

The cybersecurity department defends the federal government’s national security and the country’s military-industrial base. This is the highest-profile department, and deliberately so. Over the last five years, the previously shadowy NSA has become more vocal and active in cybersecurity. It has launched public advisories and research projects that would once have been anathema for an organization whose existence wasn’t even acknowledged until 20 years after its founding. 

Now the products of NSA research, like Ghidra, a free, sophisticated reverse engineering tool that helps in the technical dissection of hacking tools, as well as other software, are popular, trusted, and in use around the world. They serve as powerful cybersecurity tools, a recruiting pitch, and a public relations play all wrapped into one. 

The physics department, which Herrera once directed, runs dozens of laboratories that conduct most of the work on quantum information sciences, but it has a much wider remit than that. As physical limits in the ability to squeeze more transistors into chips threaten to slow and halt 60 years of predictably rapid computing growth, its physicists are exploring new materials and novel computing architectures to drive the next generation of computing into a less predictable future, exactly the kind of task the directorate was given when it first came into existence.

Meanwhile, the electrical engineering department has been looking closely at the physics and engineering of telecommunications networks since the internet first arose. As well as the issues around 5G, it also tackles every facet of the digital world, from undersea cables to satellite communications.

Some prospects on the horizon don’t fit neatly into any particular box. The computer science department’s work on artificial intelligence and machine learning, for example, cuts across cybersecurity missions and data analysis work with the mathematicians.

Herrera repeatedly raises the prospect of the directorate needing to develop greater capabilities in and understanding of rapidly advancing fields like synthetic biology. The NSA is hardly alone in this: Chinese military leaders have called biotech a priority for national defense. 

“Much of the competition in the world now is not military,” Herrera says. “Military competition is accelerating, but there is also dissemination of other technologies, like synthetic biologies, that are frankly alarming. The role of research is to help the NSA understand what the impact of those technologies will be. How much we actually get involved, I don’t know, but these are areas we have to keep an eye on.”

Finally, the math department, the directorate’s oldest, is unique. Herrera describes math as a core defining work of the directorate. The NSA is the country’s biggest employer of mathematicians, and the directorate boasts some of the best. Virtually every other department in the NSA’s Research Directorate suffers from having to compete with tech companies and the high salaries available in the private sector. The math department does not have that issue, Herrera says. Silicon Valley typically values software developers more than it does mathematicians.

The math department, often in conjunction with the computer science department, helps tackle one of NSA’s most interesting problems: big data. Despite public reckoning over mass surveillance, NSA famously faces the challenge of collecting such extreme quantities of data that, on top of legal and ethical problems, it can be nearly impossible to sift through all of it to find everything of value. NSA views the kind of “vast access and collection” that it talks about internally as both an achievement and its own set of problems. The field of data science aims to solve them.

“Everyone thinks their data is the messiest in the world, and mine maybe is because it’s taken from people who don’t want us to have it, frankly,” said Herrera’s immediate predecessor at the NSA, the computer scientist Deborah Frincke, during a 2017 talk at Stanford. “The adversary does not speak clearly in English with nice statements into a mic and, if we can’t understand it, send us a clearer statement.”

Making sense of vast stores of unclear, often stolen data in hundreds of languages and even more technical formats remains one of the directorate’s enduring tasks.

In the digital age, one of the primary goals of spying would be the ability to decode important data are currently protected by strong encryption. That’s why the Research Directorate’s mathematicians and computer scientists design and break cryptography algorithms for some of the world’s most sensitive systems.

The building and breaking of code is at the core of what the directorate does because, when the NSA looks into the future, what it sees is an increasingly digital world filled with data. Its ability to both protect and surveil it will help define great power competition for a long time.

“In the future, superpowers will be made or broken based on the strength of their cryptanalytic programs,” a 2007 document from the agency explained. “It is the price of admission for the US to maintain unrestricted access to and use of cyberspace.”

“The Research Directorate exists to enable the mission,” Herrera says. “From atoms to systems, we do research with the mission in mind.”

Correction: We amended the description of why computing growth is slowing

Corsight AI, a facial recognition subsidiary of the Israeli AI company Cortica, purports to be devising a solution for that sort of situation by using DNA to create a model of a face that can then be run through a facial recognition system. It is a task that experts in the field regard as scientifically untenable. 

Corsight unveiled its “DNA to Face” product in a presentation by chief executive officer Robert Watts and executive vice president Ofer Ronen intended to court financiers at the Imperial Capital Investors Conference in New York City on December 15. It was part of the company’s overall product road map, which also included movement and voice recognition. The tool “constructs a physical profile by analyzing genetic material collected in a DNA sample,” according to a company slide deck viewed by surveillance research group IPVM and shared with MIT Technology Review. 

Corsight declined a request to answer questions about the presentation and its product road map. “We are not engaging with the press at the moment as the details of what we are doing are company confidential,” Watts wrote in an email. 

But marketing materials show that the company is focused on government and law enforcement applications for its technology. Its advisory board consists only of James Woolsey, a former director of the CIA, and Oliver Revell, a former assistant director of the FBI.

The science that would be needed to support such a system doesn’t yet exist, however, and experts say the product would exacerbate the ethical, privacy, and bias problems facial recognition technology already causes. More worryingly, it’s a signal of the industry’s ambitions for the future, where face detection becomes one facet of a broader effort to identify people by any available means—even inaccurate ones.

This story was jointly reported with Donald Maye of IPVM who reported that “prior to this presentation, IPVM was unaware of a company attempting to commercialize a face recognition product associated with a DNA sample.”

A checkered past

Corsight’s idea is not entirely new. Human Longevity, a “genomics-based, health intelligence” company founded by Silicon Valley celebrities Craig Venter and Peter Diamandis, claimed to have used DNA to predict faces in 2017. MIT Technology Review reported then that experts, however, were doubtful. A former employee of Human Longevity said the company can’t pick a person out of a crowd using a genome, and Yaniv Erlich, chief science officer of the genealogy platform MyHeritage, published a response laying out major flaws in the research. 

A small DNA informatics company, Parabon NanoLabs, provides law enforcement agencies with physical depictions of people derived from DNA samples through a product line called Snapshot, which includes genetic genealogy as well as 3D renderings of a face. (Parabon publishes some cases on its website with comparisons between photos of people the authorities are interested in finding and renderings the company has produced.) 

Parabon’s computer-generated composites also come with a set of phenotypic characteristics, like eye and skin color, that are given a confidence score. For example, a composite might say that there’s an 80% chance the person being sought has blue eyes. Forensic artists also amend the composites to create finalized face models that incorporate descriptions of nongenetic factors, like weight and age, whenever possible. 

Parabon’s website claims its software is helping solve an average of one case per week, and Ellen McRae Greytak, the company’s director of bioinformatics, says it has solved over 200 cases in the past seven years, though most are solved with genetic genealogy rather than composite analysis. Greytak says the company has come under criticism for not publishing its proprietary methods and data; she attributes that to a “business decision.” 

Parabon does not package face recognition AI with its phenotyping service, and it stipulates that its law enforcement clients should not use the images it generates from DNA samples as an input into face recognition systems. 

Parabon’s technology “doesn’t tell you the exact number of millimeters between the eyes or the ratio between the eyes, nose, and mouth,” Greytak says. Without that sort of precision, facial recognition algorithms cannot deliver accurate results—but deriving such precise measurements from DNA would require fundamentally new scientific discoveries, she says, and “the papers that have tried to do prediction at that level have not had a lot of success.” Greytak says Parabon only predicts the general shape of someone’s face (though the scientific feasibility of such prediction has also been questioned). 

Police have been known to run forensic sketches based on witness descriptions through facial recognition systems. A 2019 study from Georgetown Law’s Center on Privacy and Technology found that at least half a dozen police agencies in the US “permit, if not encourage” using forensic sketches, either hand drawn or computer generated, as input photos for face recognition systems. AI experts have warned that such a process likely leads to lower levels of accuracy. 

Corsight also has been criticized in the past for exaggerating the capabilities and accuracy of its face recognition system, which it calls the “most ethical facial recognition system for highly challenging conditions,” according to a slide deck presentation available online. In a technology demo for IPVM last November, Corsight CEO Watts said that Corsight’s face recognition system can “identify someone with a face mask—not just with a face mask, but with a ski mask.” IPVM reported that using Corsight’s AI on a masked face rendered a 65% confidence score, Corsight’s own measure of how likely it is that the face captured will be matched in its database, and noted that the mask is more accurately described as a balaclava or neck gaiter, as opposed to a ski mask with only mouth and eye cutouts. 

Broader issues with face recognition technology’s accuracy have been well–documented (including by MIT Technology Review). They are more pronounced when photographs are poorly lit or taken at extreme angles, and when the subjects have darker skin, are women, or are very old or very young. Privacy advocates and the public have also criticized facial recognition technology, particularly systems like Clearview AI that scrape social media as part of their matching engine. 

Law enforcement use of the technology is particularly fraught—Boston, Minneapolis, and San Francisco are among the many cities that have banned it. Amazon and Microsoft have stopped selling facial recognition products to police groups, and IBM has taken its face recognition software off the market. 

“Pseudoscience”

“The idea that you’re going to be able to create something with the level of granularity and fidelity that’s necessary to run a face match search—to me, that’s preposterous,” says Albert Fox Cahn, a civil rights lawyer and executive director of the Surveillance Technology Oversight Project, who works extensively on issues related to face recognition systems. “That is pseudoscience.”

Dzemila Sero, a researcher in the Computational Imaging Group of Centrum Wiskunde & Informatica, the national research institute for mathematics and computer science in the Netherlands, says the science to support such a system is not yet sufficiently developed, at least not publicly. Sero says the catalog of genes required to produce accurate depictions of faces from DNA samples is currently incomplete, citing Human Longevity’s 2017 study.

In addition, factors like the environment and aging have substantial effects on faces that can’t be captured through DNA phenotyping, and research has shown that individual genes don’t affect the appearance of someone’s face as much as their gender and ancestry does.  “Premature attempts to implement this technique would likely undermine trust and support for genomic research and garner no societal benefit,” she told MIT Technology Review in an email.

Sero has studied the reverse concept of Corsight’s system—“face to DNA” rather than “DNA to face”—by matching a set of 3D photographs with a DNA sample. In a paper in Nature Communications, Sero and her team reported accuracy rates between 80% to 83%. Sero says her work should not be used by prosecutors as incriminating evidence, however, and that “these methods also raise undeniable risks of further racial disparities in criminal justice that warrant caution against premature application of the techniques until proper safeguards are in place.”

Law enforcement depends on DNA data sets, predominantly the free ancestry website GEDmatch, which was instrumental in the search for the notorious “Golden State Killer.”  But even DNA sampling, once considered the only form of scientifically rigorous forensic evidence by the US National Research Council, has recently come under criticism for problems with accuracy.  

Fox Cahn, who is currently suing the New York Police Department to obtain records related to bias in its use of facial recognition technology, says the impact of Corsight’s hypothetical system would be disastrous. “Gaming out the impact this is going to have, it augments every failure case for facial recognition,” says Fox Cahn. “It’s easy to imagine how this could be used in truly frightening and Orwellian ways.”

The future of face recognition tech

Despite such concerns, the market for face recognition technology is growing, and companies are jockeying for customers. Corsight is just one of many offering photo-matching services with flashy new features, regardless of whether they’ve been shown to work. 

Many of these new products look to integrate face recognition with another form of recognition. The Russia-based facial recognition company NtechLab, for example, offers systems that identify people based on their license plates as well as facial features, and founder Artem Kuharenko told MIT Technology Review last year that its algorithms try to “extract as much information from the video stream as possible.” In these systems, facial recognition becomes just one part of an apparatus that can identify people by a range of techniques, fusing personal information across connected databases into a sort of data panopticon.

Corsight’s DNA to face system appears to be the company’s foray into building a futuristic, comprehensive surveillance package it can offer to potential buyers. But even as the market for such technologies expands, Corsight and others are at increased risk of commercializing surveillance technologies plagued by bias and inaccuracy.

Correction: An earlier version of this story said that Parabon NanoLabs has “solved over 600 cases”. The company has worked on over 600 cases, solving over 200.