Now, journalists and digital rights nonprofit organizations are pushing back against this trend and advocating for news publishers to lift their restrictions.

On Monday, Wired first reported on the publication of a new petition organized by the digital rights nonprofit Fight for the Future. The open letter does not call for any specific policy from publishers, but “applauds” the Wayback Machine for its work “at a time where many major media outlets are questioning whether to allow the Wayback Machine to continue to preserve journalism.” The petition has already been signed by over 120 journalists, including Cory Doctorow, Taylor Lorenz, and Ron Suskind.

“The Internet Archive is a national treasure. I use it daily, and have for many, many years. I cannot imagine doing the work I do without it,” MS Now host Rachel Maddow wrote in a testimonial published alongside the letter.

“The Internet Archive preserves over two decades of original reporting on music and popular culture by MTV News,” wrote Michael Alex, the founding editor of the now-shuttered music and popular culture news site. “History needs stewards. The people of the Internet Archive do an outstanding job of preserving irreplaceable work and making it available to journalists and researchers.”

PressProgress reporter Brishti Basu also signed the petition, detailing an incident when the Vancouver Police Department edited a press release after she published an article criticizing it for making misleading statements. The department then publicly accused her of falsifying information.

“I was able to use the Wayback Machine to immediately prove that the police department had changed their initial statement to make it look like I had lied in my article,” wrote Basu.

The petition follows a blog post published last month by the digital rights nonprofit Electronic Frontier Foundation (EFF), which cited Lab’s reporting. Joe Mullin, a senior policy analyst at EFF, called on new publishers to lift their limits on the Wayback Machine and instead take violating AI companies to court.

“In many cases, articles get edited, changed, or removed — sometimes openly, sometimes not. The Internet Archive often becomes the only source for seeing those changes,” wrote Mullin, noting that Wikipedia links to over 2.6 million news articles preserved by the Wayback Machine across 249 languages. “There are real disputes over AI training that must be resolved in courts. But sacrificing the public record to fight those battles would be a profound, and possibly irreversible, mistake.”

The recent rallying efforts by digital rights organizations echo public comments made by Wayback Machine’s director, Mark Graham, in the weeks after our reporting was first published. In February, Graham published an opinion piece on the tech policy blog TechDirt.

“Whatever legitimate concerns people may have about generative AI, libraries are not the problem, and blocking access to web archives is not the solution; doing so risks serious harm to the public record,” Graham said.

• A dangerous, unregulated form of gambling that allows for degenerate betting on real events, unfettered by the economic and legal rules that keep stock markets and sports betting in check, creating an opportunity for corruption and insider trading on a scale we have never seen before.
• Perfectly legal crystal balls that could replace polling and happen to come with a side of money.

Whatever they are, they’re constantly changing, and they’re increasingly becoming a part of the news business. In the last few months, Kalshi, a New York-based prediction market, has struck deals with CNBC, CNN, Fox News, and the AP, among others. Polymarket, another prediction market, announced a partnership with Substack in February and one with Dow Jones in January. Both Kalshi and Polymarket are also been positioning themselves as news providers in their own right — Polymarket, for example, borrows the language of news organizations (“BREAKING” and “JUST IN“) in its social media presence, which is dominated by tweets about the news followed by a link for users to bet on that news (and is also filled with misinformation).

Keeping up with prediction markets is practically a full-time job, and for a few journalists they’ve become an opportunity to stake out a new beat at the intersection of politics, culture, finance, technology, sports, and even possibly true crime.

“I see a lot of connections to things that I’ve covered in the past,” said Kate Knibbs, a senior writer at Wired who recently established herself as that publication’s resident prediction markets reporter. “I see it as an extension of the crypto boom. It’s a future of money story, an industry story, and very much something that is emerging as a natural extension of ongoing trends in American culture.”

Knibbs’ beat started percolating in her mind when she was on maternity leave. On the day she got back, she wrote a memo to her editors about covering prediction markets, which they were thrilled to receive — they had been talking about asking Knibbs to cover them anyway, because they knew she was interested in them.

For Knibbs, the beat is interesting not only because of potential future effects of prediction markets but also because of their ties to the past. “Someone was asking me, ‘Aren’t you worried that this is going to be like NFTs, and it’s just going to fizzle out?’ And I went, ‘Well, it literally is NFTs. It’s the same story.’ And you don’t get NFTs without Occupy Wall Street in my book. It’s all mixed together. I think we have this huge appetite for products like prediction markets because of the overall precarity of ordinary people’s finances.”

Dustin Gouker, an independent journalist who writes the prediction markets-focused newsletter Event Horizon, got interested in prediction markets because of his history covering fantasy sports and gambling. Until around 2018, he said, sports gambling involved going in person to a bookie — usually in Nevada — and placing a bet, getting a physical ticket to confirm the bet, waiting for the game to end, and then returning to the bookie to cash out any wins. The rise of sports gambling apps like FanDuel and DraftKings changed that dynamic, making gambling accessible to anyone with a smartphone. Prediction markets took that one step further, allowing people to bet on granular details of all kinds of events beyond the world of sports. “The velocity is supercharged,” Gouker told me from his office in Oregon. “You can lose a lot of money really quickly.”

Gouker writes two daily newsletters — Event Horizon and The Closing Line, about sports betting — and uses a combination of reporting methods to find stories. A friend helped him build a custom dashboard that plugs into Kalshi’s API to track trades, allowing him to quickly spot any notable movement, and he reactivated his X account, which he was on the verge of deleting, because he found much of the social chatter about prediction markets was happening on that platform.

“I wake up every day and I’m like, this is the world. I’m still in the fever dream,” Gouker told me. “It feels very Republican-coded, but that’s also because that’s why they exist, right? This would not exist without the Trump administration.” The Trump administration has sued states over their attempts to regulate prediction markets; Donald Trump Jr., the president’s son, is an advisor to Kalshi and sits on Polymarket‘s advisory board.

Gouker frequently gets texts from people within the prediction markets space about things they think he’d be interested in. His coverage is often critical (a recent newsletter was headlined “Prediction markets need to stop doing dumb shit”), but he says that’s a necessary balance to the narrative pushed by the prediction markets themselves.

“I’m a pain in the ass, but I’m a needed pain in the ass,” Gouker told me. “Am I overly harsh on them? Maybe, but I think there’s enough people glazing them out there in the world. If they do something good, I say that too.”

Knibbs, meanwhile, uses the service Unusual Whales to track particularly large movements on prediction markets, and often finds herself talking to academics and lawmakers about legal efforts to regulate them. She’s one of the few reporters on her beat (Suzy Khimm also covers prediction markets for NBC News, but NBC declined to participate in this story) and she is trying to approach sourcing holistically, talking to as many people in as many fields as possible.

“I’ve been dying to report stories about how the government is approaching this, because it blows my mind that we haven’t seen anyone arrested for insider trading yet,” she said. “I think if arrests are made, we’re going to get a new level of insight into what is actually going on here, because there will be criminal complaints that we can read and hopefully things that we can FOIA. Right now, it’s pretty opaque.”

Both Gouker and Knibbs have played around with prediction markets to some extent, but neither is a big gambler. (“That’s not my vice,” Knibbs said.) Gouker has used Kalshi to place the occasional bet on college sports, and Knibbs made $50 on PredictIt when she correctly predicted that John Fetterman would be elected to the Senate in 2023. Both also think it makes sense for news organizations to use prediction market data in their reporting, as long as they’re accepted as flawed forecasting tools rather than gospel truth. But Knibbs is concerned about how prediction markets and journalism might overlap in other places.

“What is deeply disturbing to me is all of these efforts to really enmesh the prediction markets in media companies,” she said. “I’m concerned about a world in which editorial is being told that they have to use certain phrasing in order to clear up any ambiguity in how these markets resolve. That’s really gross and would be a violation of editorial independence.”

Knibbs is also on the lookout for “the first big journalist insider trading scandal.” There’s information asymmetry, she points out: Journalists often learn information before the public does, and often have off-the-record conversations that could give them insights other people might not have. They’re also underpaid, which provides them a financial incentive to act on that information. She’s not the only one concerned about that possibility; this week, ProPublica updated its code of conduct to ban journalists from placing bets on news events on prediction markets.

Gouker is less concerned than Knibbs — he points out that the history of news organizations making ad deals is similarly fraught with thorny editorial questions — but he is curious about what will happen to prediction markets if the political winds shift. Prediction markets have been positioning themselves as something akin to the news, and while that might help them build legitimacy it could also open them up to the same political attacks the news has faced since Donald Trump’s first run for the presidency back in the 2016 election.

“Will Republicans continue to lead on prediction markets when the story is not one that they want to hear or see?” Gouker asked. “If Kalshi says there’s an 85% chance the Democrats will win the House, are Republicans going to say that’s fake news? There’s this huge intersection of politics and government and tech here. At some point, does the monster start eating its own tail?”

The answer, we learned Tuesday, is no: The Venetoulis Institute for Local Journalism, the nonprofit parent organization of The Baltimore Banner, reached an agreement with Block Communications to acquire the Pittsburgh Post-Gazette, which was slated to shut down in May.

It’s a dramatic, if not entirely unpredicted, development for two news organizations whose opposite trajectories reflect some broader trends in the world of local news. The Post-Gazette is a beleaguered, historic metro daily whose union completed a divisive 1,133-day strike over health benefits last November. The NewsGuild technically won in court, but in January, Block Communications announced the newspaper’s financial losses were untenable and that it would print its final edition May 3.

The Banner, meanwhile, is a national poster child for nonprofit news success. Since its founding in 2022, when Maryland businessman and Venetoulis chairman and founder Stewart Bainum pledged $50 million over about five years to the news outlet, it has won a Pulitzer, grown into the state’s largest newsroom, and rebranded from The Baltimore Banner to The Banner, even expanding coverage into D.C.’s suburbs after Washington Post layoffs. (The Banner has not yet broken even.)

When Josh wrote about the Post-Gazette’s expected closure back in January, he observed that the Block family’s internal divisions and baggage might have stymied philanthropic support of local news in Pittsburgh. “Pittsburgh has been kind of stuck,” he wrote. “There’s pent-up capacity for something like the Lenfest Institute-era Philadelphia Inquirer or The Baltimore Banner — either a nonprofit conversion of the local daily or a robust competitor/replacement for it. But Block family drama — along with the neverending strike — complicated things enough to prevent much action.”

He even added: “The fact that the Post-Gazette announced a closure date that’s still five months off means there’ll be time for some combination of Pittsburgh’s foundations, universities, and institutions to react. Maybe that looks like the Blocks donating the Post-Gazette to a nonprofit that carries on with a decent-sized newsroom — a version of what the Salt Lake Tribune has done.”

Terms of the acquisition were not disclosed, but The New York Times reported that Bainum and his wife Sandy will commit an additional $30 million over the next five years “to help expand The Banner and turn around the Post-Gazette.” In a note to Banner subscribers Tuesday, Venetoulis Institute CEO Bob Cohn wrote that the “generous investment is designed to taper over time as revenue grows, putting us on a clear path to full financial sustainability.” The transaction will take effect May 4.

The Times also reported that Venetoulis was neither the only nor the highest bidder for The Banner — Alden Global Capital, the hedge fund known for gutting newspapers, was among the rival contenders.

“The Block family has worked to find the best possible source for responsible local journalism for the Pittsburgh region and we believe we have succeeded,” said Karen Johnese, chairperson of Block Communications, Inc, in a statement. Block Communications did not return my call for comment.

In his note to Banner subscribers, Cohn framed the acquisition as a move that lifts all boats and, specifically, “strengthens our work in Baltimore and throughout Maryland.” Spreading the cost across a broader business, he wrote, accelerates The Banner’s path to sustainability and “makes the model stronger and more durable here in Maryland as well as in Pittsburgh….Throughout this growth, our commitment to Maryland remains unchanged and central to everything we do.”

“From its launch in 2021, the Venetoulis Institute had a vision to create a nonprofit business model to address the local news crisis playing out across the country,” Banner VP of editorial and business development Monique Jones told me in an email. While that work began in Baltimore and has expanded across Maryland, the ambition of the Venetoulis Institute “has always been to take that model to other regions.” The Venetoulis team sees promise for replicating The Banner’s model in regions that “[care] deeply about the impact of local news on its communities with audiences willing to pay for news and a supportive business and philanthropic environment.”

Venetoulis expects to reach sustainability “in the next few years,” and has no plans for acquisitions or expansions beyond Maryland and Pennsylvania at this time, Jones added.

According to the Post-Gazette’s reporting, Venetoulis plans to continue the newspaper’s two print publication days; Jones confirmed those plans. While the Times reported that new ownership plans to hire back “a large number” of the Post-Gazette’s employees and run advertising and sponsorships locally, Bainum told the Post-Gazette that the “current business model does not support the size of the current newsroom,” which stands at around 100, adding, “We’re going to have to thoughtfully address that.” Jones said it’s “too early to know” how many Post-Gazette employees Venetoulis will rehire. She added that Venetoulis plans to combine back-end operations including finance, HR, subscription marketing, and technology into a shared services platform that supports sustainability across both organizations.

“Local journalism is essential to a strong community, but across the country the business model has been under severe strain,” Bainum said in a statement. “We believe there is a path forward — one that combines great journalism with a diversified business model built on scale and exceptional talent.”

“Banner leadership has already publicly indicated that it intends to cut jobs and not uphold our union contract,” the Pittsburgh NewsGuild noted in a statement. “Asset sales do not inherently get companies out from under the legal liabilities they have already incurred. The Nov. 10, 2025 U.S. 3rd Circuit Court ruling requires the company to pay back all bargaining unit employees for the costs the paper illegally passed onto them. That liability does not go away with the sale of the paper.”

“We are excited for this historic institution to survive,” the NewsGuild added, “and eager to ensure that it operates in a way that respects the people of Pittsburgh, and the journalists who strive to serve them.”

Updated April 15 with additional answers from The Banner.

But social traffic is down, too. Last week, when I was analyzing how links hurt publishers on Twitter, I asked analytics platform Chartbeat for data on how Twitter referral traffic has changed.

The decline is stark. Global Chartbeat clients’ traffic from Twitter has fallen by 70% since 2022 when Elon Musk acquired the platform.

But it’s not just Twitter. Facebook traffic has declined steeply, too.

Users are also spending less time on publishers’ sites after they click through from Facebook or Twitter. This chart looks specifically at engaged time on news and media sites:

Charts made with Claude.

To better understand the trends and challenges in the growing landscape, CNTI partnered with Project C — a research hub on creator journalism — to survey 43 independent information providers and creator-journalists in the United States. Twenty-six of the survey respondents also participated in in-depth interviews about their work. Nieman Lab readers will recognize some of the names here: Taylor Lorenz (User Mag), Kat Tenbarge (Spitfire News), Ryan Teague Beckwith (Your First Byline), and Barbara “Bob” Allen (The College Journalism Newsletter), among others.

The report, titled “U.S. Indie Info Providers: Professionally Diverse, Mission-driven, Sometimes Lonely, Rarely Earning Profit,” finds that while “indie info providers” increasingly see themselves as mission-driven small business owners, only five of the 43 respondents said they could “fully fund their lifestyle” with content creation income; just over 50% (23) said they “can’t fund their lifestyle at all” with their content. Less than one in three interviewees had a “formal or developed business strategy,” according to the report.

Like many journalists working in legacy newsrooms, news creators find their work meaningful and fulfilling, but they also worry about making ends meet and consider cash flow and managing finances to be some of their greatest challenges. Many of the creators interviewed said they rely on a mix of income sources, from freelance and consulting work to savings and support from a partner.

“Journalism isn’t immune to the larger trend of the gig-ification of labor,” CNTI senior research manager Jay Barchas-Lichtenstein told me. “Most people in the U.S. think that journalism is stably funded and that access to quality information should be a right. But instability in the industry is actually a big driver behind the indie trend. These trends are in tension: If you believe information is a public good, someone still has to pay for it. If something is valuable to you, find a way to support it financially. That’s especially true if you have the means to do so for people who don’t.”

Of the 43 survey respondents, 35 identified themselves as journalists. Many had previously worked as reporters in legacy newsrooms, others had held management positions in news, and some had no journalism experience at all. CNTI found that the news creators with only newsroom experience felt the least prepared when it came to business and operational management. Ten out of 26 interviewees had taken professional development courses to learn business skills.

Subscriptions, memberships and donations, and advertising were the most common revenue streams. Only a few respondents have found a “third pillar” to fund their content. “One sells software related to their reporting and uses some paywalled games to drive subscriptions, and the other serves as a broker for market research, connecting their professional audience to paid opportunities for a finder’s fee,” the report says.

Asking for money is also hard. Four interviewees cited imposter syndrome as a hurdle and felt as if other content is more deserving of reader revenue. Former journalists without business experience also struggle with pricing and marketing their work.

Monetizing content sometimes also conflicts with creators’ beliefs about information access. Some of the creators interviewed serve audiences that are less likely to have disposable income for a subscription publication, so they can’t rely on subscriptions or donations.

“News is so important it should not be gated…[but] news is not free to produce,” one creator said.

One of the most financially successful interviewees — whose publication serves a niche group of professionals — told CNTI that “Writing for a wealthy group of people is the only way at this point, as far as I can tell, to run a media business…90% of media businesses just write for upper-middle-class people if not just upper-class people.”

Other interesting findings from the report include:

Creators largely work alone, but rely on each other for support. Independent news and information creation is a growing field. Interviewees said they work long, intense hours, and often on their own. They look to other creators for inspiration and advice, and pay it forward when they can.

“I still dedicate a lot of time when people ask me about starting your own business or about being a solo in the newsletter world,” one creator told CNTI. “Because people did that for me and I have an ethical obligation to share that, especially now that I’ve been doing this a little bit longer.”

Maintaining a presence on multiple platforms is exhausting, but necessary. Most of the creators interviewed are active on at least three platforms to distribute and promote their work. They weigh which platforms to use based on multimedia offerings, audience preferences, and revenue potential. They described maintaining a presence on multiple platforms as time-consuming and “frustrating” but necessary so as not to become dependent on any one source for reach and revenue.

“Seeing what happened to Twitter, it was very clear to me that any tech company could implode that quickly,” one creator said.

AI has pros and cons. Some interviewees were concerned about audience use of AI as an information source. (At the same time, none of them “described using AI tools like LLMs as a distribution platform.”) Several used AI for business and production tasks, but almost none used it to actually create content. A few creators avoided AI entirely, and some were actively opposed to its use.

“I think that people should use their big brains, and you can put that in there,” one creator told CNTI.

Read the full report here.

Posts with links definitely do worse. The New York Times, which includes links in 88% of its tweets, has 53 million followers and a median of 383 engagements (likes + comments + RT’s) per tweet — an engagement rate of 0% when you calculate average engagements per follower. CNN‘s engagement rate? Also 0%. Engagement-maxing accounts like @GlobeEyeNews and @LeadingReport, which don’t include links in tweets, perform much better, with engagement rates of 0.95% and 0.45%, respectively.

But links aren’t the only thing that make or break a tweet. I used Claude to analyze the text of all the tweets in my sample and point out possible patterns. Here are a couple things that make a news tweet perform well on X.

Breaking! Across the board, tweets that begin with “Breaking” or “Breaking News:” have higher engagement. New York Times tweets that began with “Breaking News:” had an average 3,232 engagements, four times the average. A similar pattern held for tweets from the AP, Washington Post, and Wall Street Journal. (CNN rarely uses “Breaking” in tweets.)

Breaking News: Law enforcement officials are said to have disrupted a plot to assassinate Nerdeen Kiswani, the leader of one of New York’s most active pro-Palestinian protest groups. https://t.co/xWdsleQRRl

— The New York Times (@nytimes) March 27, 2026

Breaking News: The U.S. will allow a Russian oil tanker to reach Cuba, letting critical fuel in after months of what amounted to a blockade. https://t.co/FsMPauYJ72

— The New York Times (@nytimes) March 29, 2026

BREAKING: Trans women athletes are banned from the Olympics by a new IOC policy on female eligibility. https://t.co/ZgLxRn9DO9

— The Associated Press (@AP) March 26, 2026

Breaking news: Two U.S. military aircraft were shot down in separate incidents Friday while conducting combat operations against Iran, setting off a search-and-rescue effort that remains ongoing for one missing crew member, U.S. officials said.https://t.co/jiwiNNYKZr

— The Washington Post (@washingtonpost) April 3, 2026

Globe Eye News and Leading Report begin just about every tweet with “BREAKING.”

Trump quotes: Fox News’s most-engaged tweets were direct quotes from Trump, no editorializing added. Its most engaged-with tweet in my sample was his Easter message.

PRESIDENT TRUMP: "I'm proud to join with Christians across the country and around the world to celebrate the most glorious miracle in all of time: The resurrection of our Lord and Savior, Jesus Christ."

"To be a great nation, you must have religion — and you must have God. In… pic.twitter.com/1iIc1SKp4G

— Fox News (@FoxNews) April 4, 2026

~Vagueness~: Globe Eye News tweets that did not include sourcing information got nearly twice the engagement of tweets that did include a source. In other words, unattributed claims do better. Here are Globe Eye News’s most-engaged tweets in my sample:

BREAKING:

Iran announces the Strait of Hormuz is open to all countries except the United States, Israel, and their allies. pic.twitter.com/hyV49YFPs7

— Globe Eye News (@GlobeEyeNews) March 14, 2026

BREAKING:

France sold its gold stored in New York and purchased an equivalent amount in Europe.

All of France’s gold reserves are now located in Paris. pic.twitter.com/b9DgIbzZBC

— Globe Eye News (@GlobeEyeNews) April 6, 2026

Meanwhile, Globe Eye News tweets that included an “according to” — whether it was “according to” X news outlet or “according to” an investigation — were among its poorest performers.

BREAKING:

President Trump says China should help the US keep the Strait of Hormuz open, according to FT. pic.twitter.com/b8HeJPQWpT

— Globe Eye News (@GlobeEyeNews) March 16, 2026

BREAKING:

US responsible for deadly bombing of Minab school in Iran that killed 175 children and other people, according to preliminary investigation findings. pic.twitter.com/igkcetSm72

— Globe Eye News (@GlobeEyeNews) March 11, 2026

On Monday, TMZ — the American tabloid outlet known for entertainment and celebrity news — announced that its staffers are now covering Washington D.C.

“Our 3 intrepid producers — Charlie Cotton, Jacob Wasserman and Jakson Buhaj — are working The Hill,” the story reads. “So we’re in D.C. … on the hunt for good stories. We’re also going to explore the intersection between pop culture and politics. We have a lot in store in that department!”

TMZDC Staff Starts Today!!! https://t.co/KmnBdeJTaA pic.twitter.com/s7wpG8lMtW

— TMZ (@TMZ) April 13, 2026

The announcement comes after TMZ spent the last two weeks crowdsourcing photos and information to hold politicians’ feet to the fire during the longest partial government shutdown in United States history.

The Department of Homeland Security has been partially shut down since February 14. On March 26, TMZ published an interview with Rebecca Wolf, a furloughed TSA employee who was struggling to make ends meet.

TMZ was outraged, founder and executive producer Harvey Levin said in a statement to Nieman Lab, and put a call out to its audience to send in photos and sighting of politicians on spring break trips. Its following is massive; TMZ has 8.2 million followers on both Instagram and X, 6 million TikTok followers, 5.1 million YouTube subscribers, and had 47 million visitors to its website in March, according to SimilarWeb.

“We wanted to use our platforms to show how Congress — Dems AND Republicans — have betrayed us,” Levin said. “We spontaneously came up with the idea to juxtapose members of Congress on their Spring Break against federal workers who are losing their homes, their cars, their livelihoods.”

Since then, TMZ has published several stories about elected officials on vacation while unpaid DHS employees try to figure out how to pay the next month’s rent.

South Carolina senator Lindsey Graham “lives it up” in Disney World. California representative Robert Garcia was snapped at a Las Vegas Casino. New York congresswoman Alexandria Ocasio-Cortez was seen at a New York Yankees game. On March 27, President Donald Trump signed an executive order to backpay TSA workers, though the department is still partially shut down.

The photos are the paparazzi-like iPhone shots we’re used to seeing of celebrities. As Substack writer Joe Mayall tweeted, “TMZ found an interesting political niche. By covering politicians’ corruption like it’s a celebrity scandal, it attracts both audiences.”

“@TMZ has this opening because DC journalism is failing at holding the powerful in Washington to account in the Trump era,” another user tweeted.

The images seem to have resonated. I looked at TMZ’s 100 most-liked posts between March 26 and April 13. TMZ’s most-liked X post in this timeframe was a photo of Cruz on a flight out of D.C. on March 27. The post has more than 73,000 likes, 11,000 retweets, and 4.8 million views.

Senators Ted Cruz and John Thune leave D.C. amid the government shutdown.

Exclusive details: https://t.co/CUnFtLge3r pic.twitter.com/q9B9IDLAsw

— TMZ (@TMZ) March 27, 2026

Its most retweeted post is an “exclusive” story that spotted several congressional members on a trip to Scotland. For comparison, its entertainment stories from this timeframe have a median of 7,000 likes and a few hundred retweets.

Not everyone has been amused by the coverage. Capitol Hill staffers are nervous about their employers’ impending “TMZ moment,” according to Politico. On April 5, Florida senator Rick Scott tweeted a photo of himself at Disney World, saying “Hey TMZ. Yes, I’m at Disney with my grandkids. Should we be in DC? Yes! But I don’t get to make that decision.”

“Well, challenge accepted,” TMZ tweeted with its story about Scott’s post.

“Our D.C. presence will sometimes be fun,” Levin said, “sometimes intensely serious.”

About two dozen Guild members picketed ProPublica’s headquarters in New York City’s Hudson Square neighborhood during working hours, as simultaneous picket lines formed in front of the publication’s offices in Chicago and Washington D.C. On the uncharacteristically cold April morning in Manhattan, strikers bundled up in winter gear as they chanted and carried signs reading “ProPublic Workers: Deserve Fair Pay” and “Thoughts Not Bots.”

The Guild has been negotiating its first collective bargaining agreement for two and a half years, and the one-day action was intended to put new pressure on ProPublica’s management to agree to several contract proposals. The union is seeking “just cause” protections for terminations, wage increases to keep up with the rising cost of living, and contract language that would prohibit layoffs resulting from AI adoption.

“We have been trying to do this quietly at the bargaining table for two and a half years, and I’m as shocked as anybody that we are out here,” said Katie Campbell, a video journalist and member of the contract action team for the ProPublica Guild. “We need to have this done.”

The Wednesday action marked the first time a major U.S. newsroom has gone on strike, at least in part, over AI protections.

Bargaining committee members told me there has been little movement from ProPublica management since the strike authorization vote passed on March 20, with the support of 92% of the Guild. That includes the dispute over a provision that would restrict layoffs because of AI technologies. Management has offered expanded severance for AI-related layoffs as a counter proposal.

“Broadly trust in journalism is in a really fragile place,” said Campbell, noting the rise of “AI slop” and AI-generated disinformation on social media. “I would think that we would want to be leading the way on something like this. We have an opportunity to be a place that people know that they can always go to and trust that it’s going to be work that’s produced by humans.”

On social media, the Guild encouraged readers not to “cross the digital picket line” by refraining from visiting ProPublica’s website or engaging with its stories. They also asked readers not to attend a virtual event about its news app on Wednesday afternoon, which was held while workers who’d organized the event were on the picket line. A petition launched Wednesday calling for ProPublica to agree to the Guild’s contract terms had received roughly 4,200 signatures by Thursday morning.

In a statement to Nieman Lab, Tyson Evans, the chief product and brand officer at ProPublica, said, “ProPublica is committed to reaching a fair and sustainable first contract to cement the strong pay and benefits we’ve always provided our staff.” For our story on the Guild’s strike authorization vote, Evans said that ProPublica has never had a layoff in its 18-year history and that the publication is “confident we can continue to navigate future changes responsibly.”

Susan DeCarava, the president of The NewsGuild of New York, joined strikers in front of the ProPublica offices yesterday. During a spare moment on the picket line, she told me that while this strike may be setting precedent for her union, it likely won’t be the last over AI adoption in newsrooms.

“We’re going to see more and more concentrated conflicts between media bosses and journalists and media workers over who has a say and how AI is used in their workplaces,” she said.

For one, The New York Times Guild is currently in contract negotiations after its last agreement expired in February. Already, AI language has taken center stage in the Guild’s initial bargaining sessions, including over a proposal that would see Guild members receive a share of the revenue earned when their work is licensed for AI training.

During a midday rally on Wednesday, striking ProPublica employees played acoustic renditions of Bruce Springsteen’s “Dancing in the Dark” and the feminist labor song “Bread and Roses.” The backdrop for the performance: “Scabby the Rat,” the inflatable rodent used by unions across the U.S. to condemn strikebreaking activities.

New York City labor leaders from the Communication Workers of America (CWA) and the AFL-CIO addressed the crowd, as did Lily Oberstein, the chair of Business Insider’s union, another unit of The Newsguild of New York. Oberstein encouraged members to continue their fight for AI protections, pointing to Business Insider’s own layoffs of 21% of staffers last year. In a companywide memo at the time, CEO Barbara Peng said that Business Insider would be going “all-in on AI” as part of the decision.

Beyond the strike, the ProPublica Guild has also taken its dispute over newsroom AI adoption to the National Labor Relations Board (NLRB). On Monday, the Guild filed an unfair-labor-practice charge, citing a “unilateral implementation of AI policy.” The filing claims that ProPublica published AI editorial guidelines on its website last month, without first bargaining with union members over its tenets and language.

“We previewed these principles with the bargaining committee before publishing them and they offered no meaningful edits,” Evans said in a statement, calling the complaint “unfounded.”

While the dispute over AI may be the most novel part of this strike, more fundamental job protections are top of mind for some employees. That includes a provision that would require a legitimate and documented reason for firing employees, or “just cause.”

“There are people who are doing really huge investigations and award-winning work, then suddenly management’s pushing them out. That’s my biggest concern,” said Asia Fields, an engagement reporter and unit member. “ProPublica has such a great reputation — and it deserves that reputation. The journalism is so good, but I think people are surprised to know that management’s been so resistant to even basic protections.”

While having coffee with the founder of National Voter Registration Day, Montana native Matt Singer, Montana Free Press founder and executive director John Adams had a lightbulb moment: Maybe local news could benefit from that kind of national focus.

About a year later, that idea has become the first “Local News Day.” On April 9, more than 1,300 local newsrooms, around 200 partners, and 15 sponsors are participating in the “national day of action to celebrate and strengthen trusted local news and information.” (Update: As of Monday, April 13, a public dashboard and directory counted 1,462 “participating newsrooms” in the Local News Day network.)

The initiative “isn’t about fundraising specifically as much as it is about drawing the public’s attention back to local news,” Adams said. An era of constant distraction tends to “pull our attention away from local,” toward platforms and national and international news. Local News Day is designed to show the public that local news is useful and relevant to their daily lives.

The requests to the public are straightforward: “Sign up to get an email. Tune in to your local public radio station. Subscribe to a local news source,” whether for-profit or nonprofit. Regardless of business model, medium, or definition of local, “if you’re producing local journalism, we want to bring audiences to your doorstep, and that’s really what Local News Day is all about,” Adams said.

The campaign is supported by the social impact agency Impactual, where Singer is a partner. Sponsors include Press Forward, The New York Times, BlueLena, Google, and WordPress and Newspack along with their parent company Automattic. WordPress built a database of local news organizations called the “Local News Finder.” (It’s also supported by Google’s mapping technology.) The database divides outlets by audience (local, regional, and statewide) and type (newsroom, broadcast, radio, podcast, newsletter, and content creator). Newsrooms apply to be included, and some press associations have applied on behalf of their members. This isn’t the first local news mapping project, and others have learned that creating an exhaustive database of local news sources is challenging, but Adams hopes it will eventually be “one of the most comprehensive resources for locating local news in the country.”

In early discussions of Local News Day, there was some concern about redundancy or competition with NewsMatch, the end-of-year fundraising campaign for nonprofit newsrooms organized by the Institute for Nonprofit News. Adams said his vision is focused more on attention and audience-building than on fundraising. He thinks the timing can complement NewsMatch; if a community member discovers a local news organization during Local News Day, by the time NewsMatch rolls around, they might be ready to donate. (Still, if my inbox is any indication, plenty of newsrooms are making explicit fundraising asks as part of their Local News Day campaigns.)

Suggested Local News Day actions for newsrooms include launching a new product, holding an event, or dropping the paywall for a day. Montana Free Press, for instance, is asking its most loyal readers to use a unique referral link to get others to sign up for Montana Free Press; if 10 people use their link to sign up, the referrer gets a custom MFP Local News Day Yeti tumbler. If 20 sign up, they also receive a tote bag. On Thursday evening, MFP will also hold a livestreamed local event highlighting some of its reporting on how federal immigration policy has played out in rural Montana. (Meanwhile, Axios reported that Nextdoor, a Local News Day partner, launched verified accounts for local journalists. Local news was central to the company’s redesign last year.)

The long-term goal is to identify and share some best practices for audience growth from the single-day experimentation of over a thousand newsrooms, Adams said. But for the largely volunteer team behind Local News Day — many, like Adams, running their own local newsrooms while lending their time to the national project — this year, “it’s a big push to get as many people participating as we can.”

The campaign aims to generate a million news subscriptions and at least 500,000 social media follows across platforms. It’s partnering with Junkipedia, a social media analysis tool, which will help measure the reach of Local News Day across the internet, and hopes participating news organizations will share their outcomes and experiences with the organizing team.

Adams is already thinking about future Local News Days: He said he envisions more centralized partnerships between the Local News Day organizing team and institutions like libraries and universities in “Phase Two.” For now, he’s excited by the scale of Local News Day and the conversation being generated already, including county, village (!), and state-level Local News Day proclamations.

Similar to NewsMatch, Local News Day organizers are offering promotional materials and messaging tips while also leaving room for personalization. Adams said organizers took Press Forward messaging research from last year to heart and are intentionally keeping their messaging positive.

“The narrative over the last 15 [to] 20 years has been decline, desertification, distress — it’s been this really negative narrative, and that’s turned off a lot of people,” Adams said.

“Everybody knows the old, tired story of ‘news industry failing’; we don’t need to continue to talk about that,” he added. “Let’s talk about all the good things that are happening and draw the public’s attention to something they can be a part of right now to build something new and exciting.”

Back in 2016, the analytics company Parse.ly published a report: “Does Twitter matter for news sites?“

The report found that Twitter drove little traffic to most news sites, generating only around 1.5% of most publishers’ traffic. But, the authors wrote, “Twitter excels at both conversational and breaking news…Though Twitter may not be a huge overall source of traffic to news websites relative to Facebook and Google, it serves a unique place in the link economy. News really does ‘start’ on Twitter.”

Ten years later, the site formerly known as Twitter still drives very little traffic to news sites. But it’s also bad for conversational and breaking news.

Earlier this week, Nate Silver published “Social media has become a freak show.” Silver wrote that X has become “next to useless” for following breaking news like the war in Iran because its algorithm penalizes posts that include links.

“The New York Times has 53 million followers, and yet its tweets often produce only a few hundred likes, retweets, and replies even when they reveal urgent, breaking news,” Silver wrote.

The NYT published a link to critical original reporting on Iran 45 minutes ago. A good, fair story. They have 53m followers. The engagement metrics you display say they got 94 likes and 33 retweets out of that. Is that accurate? And if so, shouldn't you work on a better algo? pic.twitter.com/CMD8mfQRn7

— Nate Silver (@NateSilver538) April 6, 2026

X head of product Nikita Bier pushed back, blaming the Times’ low engagement on its paywall — but also on the quality of its tweets:

For what it’s worth:

NYT has not experimented with their captions on posts in 20 years since the launch of Twitter.

While the entire world has evolved their posting style to convert people to their newsletters (e.g., threads, etc), NYT still has their social media manager…

— Nikita Bier (@nikitabier) April 6, 2026

Big news publishers are in a strange place when it comes to X right now. Most are still posting to it (with notable exceptions like NPR and The Guardian), and they still have millions of followers. But with news business models increasingly revolving around subscriptions, publishers are focusing most of their social media efforts on sending people to their own sites. They may not see much incentive to “evolve their posting style” as Bier suggests.

I wondered, though: Is The New York Times unusual among big publishers in the “link plus sentence” tweet format? Are any major publishers moving beyond that format and seeing more engagement on X as they link less?

I used Claude to help me scrape the 200 most recent tweets from 18 large publishers’ X accounts and track the engagement (likes + comments + retweets) on each. Six of those publishers have paywalls: Bloomberg, CNN, Forbes, The New York Times, The Wall Street Journal, and The Washington Post. Nine don’t: Al Jazeera English, AP, BBC¹, Breitbart News, CBS News, Daily Wire, Fox News, NBC News, and Reuters. The last three accounts I looked at — Leading Report, unusual_whales, and Globe Eye News — are not news publishers, but aggregate breaking news in tweets without links. (Here, for example, is an example of a Leading Report tweet: “BREAKING: Iran has halted direct talks with the US, per WSJ.” They’re sometimes referred to as engagement-maxing accounts.

These charts make it pretty clear that links in tweets hurt engagement. The connection was so apparent in my analysis that a graph including all 18 publishers is almost unreadable: The traditional, link-loving publishers are clustered in the bottom left corner (lots of links, little engagement) in a nearly indistinguishable mass of bubbles, no matter how large their followings are.

If you want an easier-to-read version of the chart, here it is with some of the publishers stripped out.

Nonetheless, most of the publishers in my sample link out a lot. (I couldn’t find any biggies who are following Elon Musk’s advice to write a description of a story in the first tweet, then follow it with a link in a second tweet.) The New York Times, with 53 million followers, included links in 88% of its tweets; CNN, with 61.7 million followers, had links in 90% of tweets; The Wall Street Journal, with 21 million followers, had links in 98% of its tweets. Some examples:

Bitcoin’s founder, Satoshi Nakamoto, has remained hidden for 17 years. A trail of clues — and a year of digging by our reporter, John Carreyrou — led us to a 55-year-old computer scientist in El Salvador named Adam Back. https://t.co/s6Jy00IDdk

— The New York Times (@nytimes) April 8, 2026

(A note on the Bitcoin story: X’s Bier pointed to this tweet as a better example of how to share the Bitcoin story on X. They don’t feel meaningfully different to me, though — both are just a couple of sentences and a link. Carreyrou’s tweet, published Wednesday at 12:40 a.m., has been liked, commented on, or RT’d 9,242 times, and the Times’ tweet, published a few hours later, has been liked, commented on, or RT’d 8,361 times.)

Accused Gilgo Beach serial killer Rex Heuermann pleads guilty to the murders of eight women https://t.co/GeVNtAP68y pic.twitter.com/e8XJonGAT4

— CNN (@CNN) April 8, 2026

I checked my ego at the door. Now I’m playing faster, scoring lower and enjoying the game more. https://t.co/O8TIPMDrA6

— The Wall Street Journal (@WSJ) April 8, 2026

At the other end of the spectrum, the engagement-maxxing account Globe Eye News (with just 886,000 followers) never linked out in the tweets in my sample and saw massive engagement: A median 8,418 engagements per tweet. The New York Times, with a following more than 53 times larger, had a median 383 engagements per tweet.

BREAKING:

US demands Iran to reopen the Strait of Hormuz immediately. pic.twitter.com/pVTbzyHcQy

— Globe Eye News (@GlobeEyeNews) April 8, 2026

BREAKING: Oil prices plunge toward $90 a barrel and US stocks surge 2.7%, per AP.

— Leading Report (@LeadingReport) April 8, 2026

Trump says there may be a joint US-Iran venture for Hormuz Tolls.

— unusual_whales (@unusual_whales) April 8, 2026

Fox News (28.7 million followers) is an outlier among the traditional publishers, including links in just 9% of its tweets. Most of its tweets contain videos or graphics instead. That strategy works: Fox News had the third highest median engagement in my sample after Globe Eye News and Leading Report.

BREAKING: "From the very beginning of Operation Epic Fury, President Trump stated this would be a 4 to 6 week military operation to dismantle the military threat posed by the radical Islamic Iranian regime."

"Thanks to the unbelievable capabilities of America's war fighters, the… pic.twitter.com/d1W4G0rckb

— Fox News (@FoxNews) April 8, 2026

'GOD IS GOOD': Secretary of War Pete Hegseth gives grace to God for protecting America's heroes in the Middle East throughout Operation Epic Fury. pic.twitter.com/mU8YNIUW3H

— Fox News (@FoxNews) April 8, 2026

Links are not the only thing keeping big news publisher from high engagement on tweets; there are lots of factors. But this analysis shows that the way that most big news publishers, with the exception of Fox News, really haven’t changed the way they tweet, even as the platform’s incentives have changed.

1. Well, to clarify: The BBC recently added a paywall for U.S. users, but I looked at the BBC’s U.K. account, @bbcnews.

“The first thing newsrooms can start by doing is making it ridiculously easy for their own journalists to discover their internal archives,” Lawal said. “It’s important for the archives not to feel far off and locked away.”

Once a user-friendly internal database is in place in a newsroom, he says, journalists can be encouraged to develop new products and stories from the material.

Archives for institutional memory

Beyond editorial products, some newsrooms are also using their archives to tell the story of the newspaper itself.

Le Roch points me to an example from the French Catholic newspaper La Croix, which a few years ago published a dossier in which they used their archives to address the paper’s antisemetic past.

“I find that very interesting — a newspaper acknowledging and explaining its own history via its archives,” Le Roch said. “Not every newspaper is comfortable doing that. But a paper’s history inevitably shapes the editorial line today. When you write as a journalist, it’s your own voice, but you are also writing under the name of a newspaper with a long history.”

She suggested I speak with the team at Charlie Hebdo about this, and how they use their archives to help onboard new journalists in particular — which seemed like a good idea, because if there is any newspaper that has consequential, complicated history in French media and society, it’s Charlie Hebdo. The satirical weekly paper has long been associated with a combative, irreverent strain of French republicanism, and many people will remember it was the target of a terrorist attack in 2015 that killed 12 people.

Each of the journalists I spoke to there had a lot to say about the role of their archives in shaping their journalism.

Jean-Loup Adénor, the magazine’s deputy editor-in-chief, told me about how new team members are encouraged to spend time in the archive room, reading past issues and books about the paper’s history.

“This allows them to do two things,” says Adénor. “First, to better understand the ideological positioning of the paper and the causes Charlie has defended; and second, to draw inspiration from the tone and the freedom we have in writing here.”

“I’ve worked in more traditional media like France Info and Ouest-France, and there, it’s easy to imagine yourself in a rigid framework,” he explained. “But it’s much harder to project yourself into a free one. That’s what’s both reassuring and intimidating about the freedom here.”

“When I started at Charlie, I wanted to understand exactly what the paper was, so I spent a lot of time in the archives reading old issues,” said Yovan Simovic, a journalist who started at the paper in September 2023. “The editor-in-chief often tells us: You are completely free in your writing here. But that freedom is a bit frightening, and you don’t immediately understand what it means. So going to read the old issues, understanding how they spoke, how they described the world, how they wrote — all of that helped me to understand what he meant. Of course, we don’t want to copy [previous journalists], but it helps to see how free we are by seeing what they were able to write.”

Each of them had a favorite piece from the archive they could point to — for Simovic, it’s a piece of embedded reporting done in Afghanistan by journalist Agathe André, for Adénor it’s the survivors issue published the week after the 2015 attacks — and told me how they’d pinned articles or cartoons from past editions to the walls of the newsroom for inspiration.

Adénor suspects that understanding your newspaper’s history is particularly important at Charlie Hebdo.

“Maybe the emotional attachment to the paper is different at Charlie than at other newsrooms. I’m trying not to be too grandiloquent in what I say, but the fact is, for secular, left-wing, republican-left journalists, working at Charlie Hebdo is not trivial,” Adénor says, referring to the price many of journalists have paid (and continue to pay) to work there.

“I feel a responsibility to do my job as well as possible, and that requires knowing the paper’s history. I’m not a historian, and I don’t aim to become an expert, but I do need to understand the major milestones.”

He feels he owes it to the audience as well.

“Many of our readers have followed the paper for decades. They often know the paper better than we do. So when we receive criticism, it helps to understand where it’s coming from, historically.”

“You feel a kind of responsibility to respect that memory by understanding it,” agreed Etienne Le Page, an intern who started last autumn. “You can’t just write without context. You need to know what happened at the paper in the past, and how the paper functions and continues. That’s something you learn to do by reading past editions of the paper.”

Now, an appellate court has upheld the core findings of L.A. Press Club v. Noem and ruled in support of an injunction. In an opinion released last week, the panel of judges wrote that the plaintiffs are “likely to prevail on their First Amendment retaliation claims” because of an “avalanche” of evidence that shows DHS was acting with “retaliatory intent.” In other words, the court found that ICE officers were trying to punish the plaintiffs for exercising their First Amendment rights.
“Shooting projectiles at reporters, targeting observers, and injuring demonstrators is not crowd control. It’s retaliation,” said Matt Borden, a partner with BraunHagey & Borden LLP, a firm representing the plaintiffs, in a statement about the ruling.

That said, the judges also ruled that the original injunction was “overbroad in several respects” and sent the case back down to the lower courts in order to “fashion a narrower injunction.” In particular, the opinion took issue with the fact that the restrictions also “cover non-parties” attending protests in the Los Angeles area.

“Here’s the takeaway: Keep recording,” said Adam Rose, the deputy director of advocacy at the Freedom of the Press Foundation in an video about the ruling posted on Tuesday. “This case exists because of videos, because of photos, because people documented what was happening in real time.”

A federal court recently upheld journalists’ First Amendment rights by ruling that DHS used unlawful force against reporters at LA immigration protests last year.

Watch @adamrose.bsky.social explain, and if you’re a journalist facing a press freedom violation, contact @pressfreedomtracker.us.

[image or embed]

— Freedom of the Press Foundation (@freedom.press) April 7, 2026 at 11:50 AM

“If you are a journalist and you can’t tell the story, you are the story,” said Rose.

Among dozens of violent incidents against journalists in L.A., the original suit included evidence that ICE officers shot a journalist in the head with a rubber bullet, leading to a concussion, and hit another journalist in the arm with a tear gas canister, causing a hematoma and burns. The plaintiffs called for specific limits on ICE’s use of force, including the assault and dispersal of journalists without cause, the use of chemical and projectile weapons on members of the press, and the firing of weapons at “heads, necks, or other sensitive areas.”

In the original opinion, judge Hernán Vera wrote that ICE’s actions “undoubtedly chill the media’s efforts to cover these public events and protestors seeking to express peacefully their views on national policies.” He also said that an injunction was necessary to “curtail the federal agents’ indiscriminate use of force targeting journalists standing far from any protest activity.”

The Southern California suit has not only set precedent for restricting ICE’s use of force against the press, but also inspired similar legal actions by U.S. press organizations. Last summer, the Chicago Headline Club led a federal suit against DHS, citing incidents of violence against members of the press, elected officials, clergy, and peaceful protesters during ICE’s Operation Midway Blitz in the city. In November 2025, a judge issued a preliminary injunction limiting ICE’s use of force in Chicago. Soon after, Border Patrol Commander Gregory Bovino pulled out of the city, and the plaintiffs voluntarily dismissed their suit.

“The retreat of DHS agents from Chicago is due in large part to the bravery of our plaintiffs and every Chicagoan who spoke up about the brutality they experienced because they exercised their First Amendment rights,” said Katie Schwartzmann, counsel at Protect Democracy, in a statement at the time.

It was technology and culture reporter Taylor Lorenz who first told news creator Vitus “V” Spehar to think of themselves as a journalist.

It was 2022 and Spehar — the 43-year-old best known for their explainers as @UndertheDeskNews on social media — was two years into explaining the news online. They were interviewing Lorenz for their podcast and initially brushed it off. “No I’m not, I’m a TikToker,” Spehar said at the time.

“She was like, ‘No, and actually, that’s irresponsible for you to say. What you are doing is journalism and you need to understand the ethics and expectations. The public thinks you’re a journalist, so you need to be one,'” Spehar recalls Lorenz telling her.

Spehar’s videos — which cover U.S. politics, policy, and culture — are conversational and easy to watch. Donning their signature glasses and sometimes a suit, they don’t shy away from playing trending music, acting, or doing a little dance to help viewers understand news events (like the firing of Department of Homeland Security head Kristi Noem.)

Today, they’re one of the most successful news creators in the industry, with nearly five million followers between TikTok and Instagram. In addition to posting daily news recap videos, Spehar also streams a live newscast on YouTube, sends out a Substack newsletter to more than 184,000 subscribers (nearly 7,000 of whom are paid), interviews politicians, and appears on cable news like CNN and MSNOW. Spehar was also a 2025 fellow at the Harvard Kennedy School’s Shorenstein Center on Media, Politics, and Public Policy.

I chatted with Spehar in February about why they first started reporting from under their desk, how they stay on top of the news cycle, and the business of being a news creator. Our conversation has been edited for length and clarity.

“We’re not a newspaper company and we haven’t been for quite some time,” Julie Pace, executive editor and senior vice president of the AP, told the AP’s own David Bauder. While they once accounted for the majority of the AP’s revenue, big newspapers now only make up 10% of the organization’s income. That revenue has fallen by 25% over the past four years, in large part because Gannett and McClatchy, two of the largest newspaper companies in the U.S., stopped publishing AP wire content in 2024. According to Bauder, the AP had also learned that Lee Enterprises, another large newspaper publisher, was seeking an early exit from a contract that was set to expire at the end of the year.

Pace told Bauder that whether or not the company conducts layoffs will depend on how many people take the buyout. Still, Pace said, “The AP is not in trouble…We’re making these changes from a position of strength but we’re doing so now to recognize our changing customer base.” The AP will be upping its video teams, as well as adding journalists to beats “on topics of known customer interest.” It will also still have journalists in all 50 states.

Lately, the AP has been looking to tech companies for revenue (and says its revenue from such deals has grown by 200% over the last four years). It’s made deals with Google and OpenAI, and in March announced a deal to provide election data to the prediction market Kalshi. Elections in particular are a big money-maker for the AP; last year ABC, CBS, NBC, and CNN all signed up for its service providing election data.

According to the AP News Guild, the union representing AP staff, 120 people were offered buyouts. The union also said that the AP ignored a union request to bargain over artificial intelligence.

“AP continues to get rid of experienced staff and flirt with artificial intelligence” the union said in its statement, “ignoring the opportunity to differentiate AP stories as ones that are and always will be created by human journalists.”

Statement from on #AP buyouts today from the union's executive committee. @NewsMediaGuild pic.twitter.com/jB9DICuNiD

— AP News Guild (@APNewsGuild) April 6, 2026

I got my start in journalism by writing stories for the Associated Press in Curaçao.

Really sad to see the abandonment of reporting and the written word.

[image or embed]

— Karen Attiah (@karenattiah.bsky.social) April 6, 2026 at 1:43 PM

The Associated Press is pivoting away from… journalism. We’re well and truly cooked folks.

[image or embed]

— Meghan Herbst (@megherbst.bsky.social) April 6, 2026 at 1:40 PM

uh oh what are they gonna put in the w*shington p*st … https://t.co/pIxTb1sFia

— Caroline O'Donovan (@ceodonovan) April 6, 2026

Well, as they say on @LinkedIn, I might be really Open to Work soon! I'm still loving what I'm doing at AP, but new challenges are fun too. If you know of something that might align with my skills and interests, hit me up!https://t.co/Q1v4rKp35X

— Mike Sisak (@mikesisak) April 6, 2026

Massachusetts has some of the highest housing costs in the country. The problem is exacerbated on Cape Cod’s Outer Cape, a region that includes Provincetown, Truro, Wellfleet, and Eastham. In a community where the same one-bedroom apartment can be rented out for $2,500 per month year-round or $5,000 a week in the summer, the incentives for short-term rentals are overwhelming, leaving few to no options for young reporters whose salaries start at $45,000 a year.

This reality makes it challenging for the Provincetown Independent to attract the early-career reporters it wants to employ and train. The local weekly newspaper founded in 2019 is a for-profit, public-benefit corporation supported by a parallel nonprofit, the Local Journalism Project. Core to that nonprofit’s mission: educating, training, and financially supporting aspiring reporters. But how do you educate and train young reporters if they can’t find a place to live?

Essentially, you can’t, which is what the Local Journalism Project’s board has been learning the hard way. “We had raised money for young journalists; we had money sitting in the bank [for] young journalists,” board president Willow Shire told me. While they receive an abundance of “stellar applicants,” Shire said, without a place to live, “all of these candidates were turning [the Independent] down.” For the few who did choose to come to the Cape, the seasonal rental market meant they generally had to move multiple times a year. “We didn’t anticipate [housing] being as big a problem as it turned out to be,” Provincetown Independent co-founder and editor Ed Miller said.

Last year, the LJP board decided it needed to take a big swing. The nonprofit launched a capital campaign to raise $500,000 with the goal of purchasing a condominium to house three early-career Provincetown Independent reporters.

It exceeded its goal in two months, and purchased the three-bed, three-bath property in October for just under $1.5 million, making a cash down payment of $448,500 and taking out a $1,046,500 mortgage. Three reporters moved in within weeks; they pay 30% of their salaries as rent, well below market rate.

More than a century after the heyday of company towns, employer-owned and assisted housing is part of the policy landscape in other communities where people struggle to work where they live. On the Outer Cape, from restaurants to healthcare, “housing is a conversation for every business out here,” said Local Journalism Project executive director Janet Lesniak. The Local Journalism Project had previously taken a page from the Cape Cod Baseball League’s book to set up homestays for its seasonal fellows. Other newspapers in the broader region are also getting proactive about providing housing for their reporters; the Vineyard Gazette, on Martha’s Vineyard, advertises its new fellowship as including housing in “a furnished, well-appointed two-bedroom guest house.” But taking out a mortgage to buy a standalone condo is a strikingly bold investment in making local journalism accessible to young reporters.

“We did not expect to be in the housing business; we didn’t particularly want to be in the housing business,” Shire said. “But it is in the critical path of the mission. So here we are…We determined the need, we raised the funds, we found the properties, and we are teaching young journalists what a circuit breaker is.”

What makes a successful capital campaign?

The capital campaign brought in donations from close to 100 local individuals and family foundations; gifts ranged from $10 to $150,000. On top of reaching its goal — thanks in part to large donations from board members — the campaign attracted “a significant number of new donors,” Lesniak said. Some donors were longtime subscribers or seasonal readers, and others were inspired by the mission of educating and investing in young journalists. And, Miller said, “there’s a certain kind of donor who really prefers to give money for something concrete, brick and mortar.”

To Miller, the campaign’s success was, in part, a testament to the Indie and LJP’s years of work communicating the importance of local journalism, and explaining their own unusual structure. After six years operating the newspaper as a hybrid for-profit in relationship with a separate nonprofit, the community had gotten used to both paying for the newspaper via subscriptions or newsstand purchases and supporting its public service function with donations to the nonprofit. “In the early years, it was sometimes hard to explain to people” why the newspaper was asking for both sources of support, he recalled. “But we don’t get those questions anymore. I think people really get it.”

“We were stunned” by the success of the capital campaign, Miller said. But he also attributed its success to a shared understanding of the challenge of year-round housing for young people on the Cape, which “has been kind of the number-one topic of concern and civic debate for a while now — the whole community is really aware of what a problem it is and how important it is to have younger people able to live here and work here.”

“So the project made sense to people,” he said. (It occurred to me that this campaign, by tapping into such a relatable local issue, maybe even humanized the Indie’s journalists.)

One (home) leads to another

Not only did the Local Journalism Project reach its fundraising goal and purchase the condo; a community member (and friend of Shire’s) has since agreed to donate an additional three-bedroom home in Eastham to the nonprofit.

While this home needs to be gutted and renovated after not being maintained for the last 20 to 25 years, Shire is taking advantage of local community networks and philanthropic spirit to complete the renovations for “an amazingly small amount of money,” which she said has already been raised (she expects the total price tag to come in under $100,000, including a new roof).

Specifically, Shire’s partner, Jaime DaLomba, is serving as the pro bono project manager; she said he’s been working on the renovations seven days per week. A welder born in Provincetown, DaLomba has leveraged the depth and breadth of his connections with other local craftspeople to convince others to donate time and basics like paint. The renovations only began in January, but Shire said progress is on track to have the home move-in ready by June 1.

“You need to be embedded in your community at all levels — we’re embedded with the local tradespeople, the lobstermen, the local painters, the local Ace Hardware store owner,” Shire said. “Your donors are critical — we have two really big donors, and we couldn’t do it without them. But we also couldn’t do it without Jaime, who quit school in 10th grade and is a welder. So you need everybody.”

“Very lucky timing”

Jack Styler‘s path to the Provincetown Independent was, in some ways, unlikely. He’s from Milwaukee, and did not study journalism. But after graduating he moved to Kazakhstan, where he did some blogging. Clips from that blog helped him get an internship at The American Prospect in Washington, D.C., and an editor there helped connect him with the Provincetown Independent. He joined as a staff reporter in June of 2024. Last month, his stories covered everything from Flock camera surveillance to a Pedicab cap bylaw to the deaths of two fishermen after a scallop boat capsized.

On the Cape, Styler’s first living arrangement consisted of sharing the second floor of a house with two other reporters; the first floor was a short-term rental, but the Local Journalism Project leased the second floor, three-bedroom apartment “on a longer-term basis.” He left the Indie for a yearlong Fulbright in Latvia, but returned as a reporter last summer. Between August and October, he bounced from a room in a house in Wellfleet, where he lived with a retired couple, to a guest house available for about a month in the same town. Then the Local Journalism Project closed on the condo.

“I moved in right away, because I was going to be without housing,” Styler said — he needed to move out of the guest house the same week the Local Journalism Project closed. “So it was very lucky timing.” He moved in with no furniture, and slept on an air mattress. It’s his first time living in a house with his own bathroom.

I commented that bouncing around that frequently sounded exhausting. “I think that out here, it just is the reality for so many people that it wasn’t an unexpected thing that I would have to jump between a few housing situations before getting permanent housing,” he said. “That’s pretty typical out here…Housing is probably the number one issue for people of all types, all walks of life on the Outer Cape. Young reporters at the Indie have definitely experienced that as well.”

Two other reporters, his roommates, moved in a couple weeks later — one a new staffer, the other someone he’d overlapped with in the summer of 2024. They were able to furnish the house quickly with donations.

Living in the condo has put the reporters, board members, and other Indie staff on intimate terms. For instance, when the lock on the bathroom door didn’t work, Styler told me Shire helped teach him to reattach the strike plate so it could lock.

“That was the moment I realized, ‘oh, right, we actually own this house,'” Styler said. “We’re able to make normal renovations, easy renovations, ourselves. And that felt kind of empowering, because…Instead of a faceless corporate entity, like so many landlords I’ve rented from, these are people who are doing this because they want to support the work that we do, and are also being helpful in a million different ways.”

Indeed, board members are on call for everything from installing a mailbox to troubleshooting a plugged drain, Shire told me. (She confirmed she showed Styler how to use a wood chisel and screw gun to fix that bathroom door; he emailed her when he successfully used the screw gun to fix the rickety top of a table missing its screws.) Styler, his roommates, and Indie staff worked together to install blinds for the first time.

The home, Styler said, is in the heart of Provincetown. That’s ideal for the local reporters.

“I think it’s super important for us to live where we live, which is right smack in the middle of Provincetown,” he said. The town “can get pretty quiet during the winter, but if there’s anything going on, our house is both not super loud, but pretty much in the middle of things, which is really, really helpful for a local reporter.”

The condo purchase, Styler added, helps him “envision a longer-term tenure at the Indie.” If you’re moving at least twice a year, he reflected, every time you move is a natural “hinge point” where you might question whether the lifestyle is sustainable and consider not just moving elsewhere on the Cape, but pursuing a different opportunity.

“I think that the idea that the Local Journalism Project actually owns this house, and that as long as I continue to do good work as a journalist and they’re happy with what I’m doing, that I should have stable housing here, is a real relief,” he said, “and kind of takes away those hinge points where you’re naturally looking for other opportunities to work other places or move other places.”

The new four-year agreement guarantees a 32.5% increase to the minimum salary for reporters, more paid time off, and journalists’ input on the use of artificial intelligence. Here’s what the contract announcement says about AI:

Provisions on use of generative artificial intelligence that include:

• At least 60 days notice to the Guild of intention to use a new generative AI system that will have a meaningful impact on terms and conditions of employment of bargaining unit employees in their performance of their work
• The ability for the Guild to negotiate effects of AI introduction, enhanced severance of four additional weeks per year of service (and a minimum of 12 weeks) for layoffs directly and primarily due to the use of generative artificial intelligence
• Ability to withhold byline or raise ethical objection to use of AI in an employee’s work
• The creation of a committee made up of Guild members and VTDigger staff members who are not in the Guild to make recommendations for the organization’s AI usage policy. The policy will include an editorial review process to determine which editorial content is subject to the policy and an acknowledgement that generative AI tools do not adequately substitute for human judgment in the creation, distribution and promotion of journalism.

A story published Thursday by The Boston Globe reports that VTDigger is struggling. CEO Sky Barsch is leaving after three years on the job, along with editor-in-chief Geeta Anand, who joined last year. VTDigger brought in $2.7 million in revenue in 2024, and increased that number by roughly 10% in 2025, Globe media reporter Aidan Ryan writes, but ” has been operating in the red since just before its founder left in 2022, and the chief executive who succeeded her is now leaving. Meanwhile, VTDigger is in the market for its third top editor in just over a year after a bruising contract negotiation with the newsroom union.” VT Digger has more than 9,000 paying members.

“The current mishegas, stemming from acrimonious contract negotiations with Digger’s union over the usual (pay and benefits) as well as how AI will and won’t be used, is a consequence of what sounds like a poisonous relationship between the two sides,” media analyst Dan Kennedy wrote Thursday.

While Barsch told Ryan she isn’t leaving because of what happened at the bargaining table, “there were hard moments.” Meanwhile, the piece cites “a number of reasons for [Anand’s] departure, including the challenging contract negotiations, Barsch’s decision to leave, and a health issue that she had been dealing with.”

Digger founder Anne Galloway — who left the nonprofit in 2022 — was more candid. “If the guild continues to be unreasonable like this, news organizations like Digger will go out of business,” Galloway told Ryan. “It’s going to be hard to attract good leadership with the kind of antagonisms that have become so endemic to the organization.”

According to the Globe, negotiations became tense when the guild started publicly campaigning for AI protections. Per Ryan:

Amid the fight, a Reddit post called on people to “target” Barsch, editor Geeta Anand, and VTDigger board members, according to Galloway and Kevin Ellis, a former Vermont Journalism Trust member.

“When you’re using that kind of language in a Trump environment, that’s frightening,” Ellis said in an interview. (The word “target” was later changed, Vermont’s Seven Days reported.)

Norm Welsh, administrator of the Providence News Guild, which represents unionized VTDigger employees, maintained that the talks were “relatively smooth.”

“I think it was a normal course of negotiations,” Welsh said. “I don’t think anything was meant personally.”

The VTDigger Guild is the latest newsroom union to push back on companies’ embrace of AI. Last month, the ProPublica Guild voted to authorize the first U.S. newsroom strike over AI protections. In November, an arbitrator found that Politico violated its contract with its unionized journalists when it rolled out two AI-powered editorial products. As of last month, 58 newsroom unions under the NewsGuild had some form of AI protections in their contracts.

Read the full story in The Boston Globe here.

That’s an amount that few people are willing to pay. There are some initiatives that do things differently, like the publication Flaming Hydra, which has 65 contributing members — and looks more and more like a magazine, complete with print editions for top-tier subscribers — and the app Noosphere, which Hanaa’ wrote about last year. But the true holy grail of newsletters, the thing that would give readers the closest thing to a bespoke magazine without making them pay full price for a bunch of individual subscriptions, is the bundle.

“Whoever figures out a way to bundle independent journalism subscriptions will be a hero,” independent journalist Marisa Kabas, author of The Handbasket, wrote on Bluesky in February.

Last week, Kabas followed up with an announcement. “Exciting news,” she wrote. “We’ve finally found a way to do a paid indie media bundle so you don’t have to separately subscribe to so many newsletters.”

Kabas had teamed up with Katelyn Burns, author of Burns Notice, and Kat Tenbarge, author of Spitfire News, to offer a 30-day bundle of all three newsletters, allowing readers to access them with a one-time $8.50 transaction — half the price of paying for all three individually.

The idea had been percolating for a while. A few months earlier, Kabas had heard from Michaël Jarjour, an ex-Twitter partner manager and co-founder of Trustfnd, a new service that allows independent journalists to create newsletter bundles and tap into each other’s audience bases. “I wasn’t able to give the idea much energy while I focused on my work and business,” Kabas told me in an email, but it was certainly something she was interested in. The discussion was revived in February, when Burns told her she’d been talking to Jarjour about launching the first paid subscription bundle. Burns, Kabas told me, was “foundational to making this happen.”

Trustfnd (pronounced “trust fund”) solved a key technical problem for Kabas, Burns, and Tenbarge: Their publishing platforms (The Handbasket and Spitfire News are on Beehiiv, and Burns Notice is powered by Ghost) don’t have a built-in way to create a bundle of any sort, whether cross-platform or on a single platform. This is true across newsletter services; Substack and Patreon also do not offer bundles.

This is partly by design. “We’ve always talked about doing this, but it gets pretty messy and complicated if the entities in the bundle aren’t actually a part of the same company,” Tyler Denk, CEO of Beehiiv, told me in an email. He outlined some potential concerns:

• What if person A drives 10x more signups than person B and person C, do they all split evenly?
• What if person C decides they no longer want to be a part of the bundle, do they take those subscriptions with them? If so, what price do they pay?
• If a reader subscribes to the bundle but primarily engages with only one newsletter, who “owns” that subscriber for purposes of future direct communication, re-engagement campaigns, or list sales?
• What if person C grows dramatically mid-bundle and wants to reprice their standalone? They’re now locked into a bundle price that undervalues them.
• What if one newsletter in the bundle publishes something controversial that creates reputational blowback for the others?

“There’s a bunch of other complications, but the concept itself I think is interesting,” he said. “It’s worth thinking on a bit more, but I do think a lot of these people are going to run into future issues with the bundle.”

Alex Kisielewski, Vice President of Partnerships and Business Development at Ghost, told me over email that bundles are “definitely on our radar,” and that Ghost has seen an uptick in requests for bundle support over the last six months. “Independent journalists are looking for ways to collaborate more, whether that be sharing audiences, co-publishing, or pooling resources,” he continued, but “there’s no getting around the fact that it’s complicated.” He echoed some of Denk’s concerns about billing and subscriber management, as well as technical challenges around access control. “We’re watching this space, and I’m glad the folks at Trustfnd are building specifically in this space,” he said.

Trustfnd works by tapping into Ghost and Beehiiv’s APIs. Independent journalists connect their newsletters to their Trustfnd accounts, and then connect their Trustfnd account with the accounts of whomever they want to team up with for a bundle. “It’s like a network effect as a service,” Jarjour told me over email. “We want them to stay independent entities but move like one when it comes to growing their owned audience.”

Bundling, Jarjour told me, allows newsletters to grow faster and more cheaply because each newsletter can tap into a shared audience funnel; readers who subscribe to The Handbasket, for example, will now be exposed to Burns Notice and Spitfire News. (Existing subscribers of each newsletter also get discounts on the bundle, adjusted according to which newsletter — each of which has a different individual subscription price — they subscribe to.)

“Turning followers (you rent) into subscribers (you own) is a collective challenge for journalism,” Jarjour wrote. “That’s why I felt the solution needs to have a collaborative element as well.”

And, he said, “I had some money on the side because I got fired from a fancy tech job by a lunatic who hates journalism [Jarjour left Twitter in February 2023, five months after Elon Musk took over]. So I wanted to build something I believe will make journalism stronger against even its worst enemies.”

Ghost and Beehiiv are the only platforms that currently support paid bundles — Trustfnd currently offers 30-, 60-, or 90-day bundles, and the Kabas/Burns/Tenbarge bundle is currently just for the first month, although Kabas said she would “definitely be supportive of a year-long trial.” Ghost and Beehiiv were easy to integrate with because they are open platforms, but Jarjour said he and his cofounder, René Pfitzner, who was previously the CEO of an ecommerce platform, are starting talks with closed platforms (such as Substack and Patreon) to integrate with them as well.

Trustfnd is currently in beta, and free for journalists to use, but plans to charge a fee (which they are currently trying to figure out) rather than take a cut of revenue. In the near term, Jarjour said he hopes to grow Trustfnd by building out bundles and creating a service for “legacy brands to get in on the action by building networks of independent journalists around their brands.”

“Our longer term vision is to enable independent entities to move like one when that is useful,” He continued. “So they can grow, earn, and spend money together. Like a new kind of news organization.”

This story has been updated to include comment from Ghost.

“Starting in May, all newly published stories on sltrib.com and in the app will be free to read — no subscription required,” wrote CEO and executive editor Lauren Gustus in a note to subscribers.

We’ve been telling you about this shift for a while now — this Sarah story from September, this Sophie story from October, and this Hanaa’ story from November. The Tribune is in a unique position among American newspapers, having converted to nonprofit status in 2019. In the years since, it’s achieved financial stability and had the space to think about some foundational questions: What should a nonprofit newspaper look like? What does it owe to a community that a for-profit might not?

One answer: It wouldn’t limit access to quality local news to people willing or able to pay. “We believe trusted, independent journalism is a right — not a luxury,” Gustus writes. “And at a time when misinformation spreads faster than ever, expanding access isn’t just important — it’s necessary.”

Here’s how it will work: When the switch is pulled in May, current digital subscribers will be converted to monthly donors at the same level. (The base rate for a digital subscription is currently $10/month.) The paper says it’ll make it easy for current subscribers to opt out, but I’m guessing most people will keep paying — hey, now it’ll be tax-deductible.

Those who transition to regular donors will get to keep a few exclusive benefits: the ability to comment on stories and the e-edition (still a surprisingly powerful draw for many iPad-wielding ex-print subscribers). They’ll also have exclusive access to older stories; it’s only newly published stories from May on that will be free to all.

How will they pull it off? In 2025, the Tribune made about $2.6 million in digital subscription revenue — about 20% of total revenue. That’s about to go to $0. The goal is to make that up with additional donations, which will obviously depend on how many current subscribers will be happy to continue being charged for a (mostly) free product. A limited test the paper ran found that 87% of subscribers stuck around as donors after being told they’d be getting the news for free. (The Tribune received a $1 million pledge from a local couple for the transition, with a 1.5× match that has been met.)

As a nonprofit, The Salt Lake Tribune is in a meaningfully different position than most American newspapers. But I think there’s a potential lesson for for-profit papers, too. At this late stage in the decline of the newspaper business model, a lot of digital subscribers are paying for civic reasons at least as much as practical ones. Salt Lake City’s metro area, defined broadly, has about 2.7 million people. As of last fall, the Tribune had 32,166 digital subscribers and 7,716 print subscribers — only about 1.5% of the market. It’s hard to build a “mass” media company on those numbers, but they can support a civic institution. Patronage isn’t just for Patreon.

Press Gazette reported that the move is part of the publication’s multi-year transformation plan, Project Berger, which aims to make The Guardian “more visual, digital, and experimental.”

Feast is one of almost 60 newsletter offerings from The Guardian, per Press Gazette. The food newsletter has more than 100,000 subscribers and open rates close to 70%.

The Guardian’s head of newsletters, Toby Moses, told Press Gazette that he saw food reporting as a natural starting point for The Guardian’s Substack foray given the strength of its own coverage and the platform’s “thriving” food scene. Some of The Guardian’s individual food writers already have their own Substack newsletters.

The Guardian is just the latest publication to experiment with Substack; The Financial Times and The Economist recently launched Substack newsletters, Press Gazette notes. New York Magazine began publishing on Substack last spring; Consumer Reports revived its 1940s-era print newsletter on the platform last November.

Read more in Press Gazette’s story here.

The site’s most successful stories in January were a very Colorado mix: a dispute over clearing forest land to build a road to 19 proposed luxury homes; a solo hiker killed by two yearling mountain lions; an Air Force captain setting a skydiving-in-formation world record (104 people!); and a local lawyer’s proposal to rename dozens of Colorado mountains. (Many of them to light-related names — ROYGBIV Peak! — and seven to each of the deadly sins. Damned Colorado bureaucrats, denying the people the chance to climb Lust Peak.)

This is the latest installment in our regular look at web traffic at the nation’s nonprofit news sites — a group which, for our purposes, is defined as the nonprofit members of the trade groups LION Publishers and the Institute for Nonprofit News.

January was a good month; the 100 most-popular nonprofit news sites totaled 72.1 million visits in January, up from 64.6 million in December. 70 of December’s top 100 saw traffic increase in January. Some of the biggest gainers included Columbus’ Matter News (up 297%), Illinois’ Lansing Journal (up 261%), Sahan Journal (up 181%; it was also the biggest traffic gainer in Q4 2025), Oaklandside (up 115%), Mississippi Today (up 113%), and The Texas Observer (up 114%)

Here’s January’s Top 25, followed by some additional detail on what worked well that month for The War Horse (up 238% over December), National Parks Traveler (up 75%), Votebeat (up 236%), and inewsource (up 211%).

The War Horse

The War Horse, which covers the U.S. military and veterans’ issues, saw its traffic jump 238% from December to January, from about 90,000 visits to 306,000. Hrisanthi Pickett, the site’s audience engagement editor, attributed the increase to a January 14 personal essay by an Army colonel reflecting on his retirement. (“The morning after my retirement ceremony, I opened the closet and stared…Who am I now, without the uniform?”)

“We did not do anything different when it came to publishing or promoting the story,” Pickett said. “The story found its own audience in private messaging groups (i.e., WhatsApp, Telegram). This was a bit frustrating to track.” The traffic uptick lasted about a week, and while it led to new newsletter subscribers and followers on social, the relative opacity of messaging platforms has made it difficult to replicate the success, she said.

National Parks Traveler

January traffic to National Parks Traveler was up 76% month-over-month, jumping from 113,000 to 198,000 visits. Editor-in-chief Kurt Repanshek said, unsurprisingly, that the Trump administration’s impacts on the National Park Service have been a major driver of interest. Two January stories stood out. One was this piece on an internal NPS email saying that placing a sticker over Trump’s face on your National Parks pass could invalidate it.

The other was this piece on Donna Shaver, one of the world’s preeminent experts on sea turtle conservation, being forced to resign her position at the Padre Island National Seashore after being critical of cuts to conservation programs. “This has been an ongoing story for some years, and there is great interest in how Shaver is being treated by the NPS,” Repanshek said. “I believe the story broke on a Saturday morning, and by Monday it had been viewed by more than 100,000.”

Votebeat

Votebeat had a banner January, with visits spiking 236% from December, from roughly 21,000 to 72,000 visits. (Those Similarweb estimates line up pretty well with Votebeat’s internal visits numbers, which were 22,957 and 88,903, respectively.) Lauren Aguirre, the site’s growth and community editor, pointed to this piece on a Wisconsin court battle over absentee voting (a hit on Google Discover) and this one on whether Trump can cancel the midterms (did well in search). About 30% of Votebeat’s traffic now comes from search, “a meaningful increase compared with our early years as a newsroom,” she told me.

inewsource

inewsource, the stalwart (if frustratingly uncapitalized) local news site in San Diego, saw its visits increase about 211% in January, going from about 36,000 to 111,000 in Similarweb’s estimate. Giovanni Moujaes, the site’s assistant editor for audience and innovations, said it’s seen “a pretty considerable traffic boost from Google Discover and other Google properties” over the past year. One area they’re investing in: Nextdoor. (“While not a large source of web audience yet, our Nextdoor community continues to grow, with two of our reporters regularly engaging with users on the platform.”)

In a new “policy advisory opinion,” the Board expressed concerns about how effective Community Notes would be in a litany of circumstances, “including in repressive human rights regimes, in particular electoral contexts and in ongoing crisis and conflict situations.” Overall, the Board warned that expanding Community Notes outside the U.S. could “pose significant human rights risks and contribute to tangible harms that Meta has a responsibility to avoid or remedy.”

In January 2025, Meta announced it was getting rid of its fact-checking program in the U.S. Launched a decade ago, the program relied on a network of third-party fact-checkers to verify content and flag disinformation. These partnerships with news and civil society organizations have been essential to the platform’s moderation practices on Meta, Instagram, and Threads.

In place of proper fact-checkers, Meta has rolled out Community Notes in the U.S., which rely on crowdsourced, user-generated footnotes to label content that is false or misleading. The decision coincided with the beginning of the second Trump administration and an explosion of AI-generated imagery on Meta’s platforms.

While Community Notes started stateside, the company also announced last year that it has plans to expand the program globally. That’s where the Oversight Board has stepped in. The Board was asked by Meta last fall to review whether certain countries or territories should be omitted from the expansion, and issue more general recommendations on the global rollout.

In addition to cautioning Meta from using Community Notes in the many countries worldwide that meet its criteria for concern, the Board also outlined several structural problems in the Community Notes model. For one, it creates little incentive not to post false or misleading content. The Board found in its review that there are “no strikes for posting content that receives a community note,” and more tellingly, there are no punitive effects on reach or monetization of posts.

The Board also found that crowdsourcing moderation would inevitably “privilege dominant political, ethnic, or minority groups.” It’s a concern that should not be taken lightly given the company’s past complicity in the genocide of minority groups in Myanmar and Ethiopia through failures to moderate hateful content. Back in 2018, Facebook apologized for its role in “offline violence” in Myanmar.

“Delays in note publication, the limited number of published notes and its dependence on the broader information environment’s reliability raise serious doubts about the extent to which community notes can meaningfully address misinformation linked to harm,” wrote the board.

The Board stopped short of recommending that Meta end the Community Notes program entirely, instead writing that more “sufficient testing and detailed data” would be required for an evaluation. Regardless, it is worth noting that Meta is not legally required to comply with any of the Board’s recommendations, and that it has only implemented them in about 75% of cases.

The Board’s opinion has already put some renewed pressure on Meta to halt its global Community Notes rollout, and even to restore its fact-checking program in the U.S.

“The Oversight Board advises Meta not to expand community notes in countries and contexts that are particularly fraught, because community notes can be manipulated by large groups,” Angie Drobnic Holan, the director of the International Fact-Checking Network (and a 2023 Nieman Fellow) wrote in a post on Poynter. “Isn’t the United States today a place experiencing one or more of those conditions?”

In the first six months of its U.S. rollout, Meta’s chief information security officers said that its platforms have only published 900 Community Notes. By comparison, Holan notes that in the same period across the E.U., professional fact-checkers helped Meta apply labels on roughly 35 million Facebook posts.

“Meta should restore professional fact-checking for the benefit of the American public, and it should do it before the next election cycle,” she wrote.

You can read the full Oversight Board advisory opinion here.

Most of the people that most news organizations are trying to reach, however, are not news junkies. Nieman Lab founder Josh Benton likes to remind us that the average American received zero push notifications from a news organization in the last week. Only 16% of Americans pay for news.

So we often discuss how we might better see beyond our own news habits. In this new occasional series, “News Diaries,” we’ll talk with people who are not professional journalists about their habits for finding and consuming news, and what counts as news to them. We’re interested in the ways news does, or doesn’t, fit into people’s lives, and the roles traditional news outlets, informal networks, and everything in between play in keeping people informed.

For the first entry in this series, I spoke with (a different) Sophie, 37. She’s a minister and mom of two who lives in Minneapolis, her hometown. We talked in mid-February, toward the end of the Trump administration’s “Operation Metro Surge,” which sent more than 3,000 federal agents into Minnesota to arrest undocumented immigrants; agents killed two American citizens.

My conversation with Sophie, lightly edited for length and clarity, is below.

One might see this news and wonder about Brittin’s editorial experience. He has very little but, as both the Guardian and the Telegraph took pains to point out, he is a Doctor Who fan, which I guess means he does at least like one BBC property. The Guardian reports that the BBC plans to hire a deputy director general to support Brittin, and that “that figure is expected to have serious editorial experience.”

Among other things, Brittin is going to have to contend with a $10 billion lawsuit from President Trump over edited video clips in a documentary, a scandal that forced out Brittin’s predecessor, Tim Davie. The BBC is also contending with dwindling audience numbers, even as it remains one of the largest public broadcasters in the world, and its royal charter — essentially the agreement between the BBC and the British government that lays out its mission and public obligations — is up for renewal in 2027. The hope seems to be that Brittin’s experience dealing with governments during his time at Google will help with negotiations, and that his tech background might help the BBC, which announced a deal with YouTube in January, regain audiences that have drifted towards social video.

Brittin was also a member of Great Britain’s 1988 Olympic rowing team, which delightfully means that British Rowing had an excuse to publish an in-depth profile of his rowing career. He appears to be a fan of AI; the Guardian writes that he “has previously urged the TV industry to ‘leap and to learn’ from the technology.”

The more the editors looked into Tom, the more alarmed they became. The bot made decisions on its own and even exchanged messages with them. “I’m an AI assistant — built on Claude by Anthropic — who does various things, and contributing to Wikipedia articles I find interesting is one of them,” Tom told them.

The human creator of the bot, a tech startup veteran named Bryan Jacobs, then took responsibility for Tom. In his first extended interview, Jacobs talked with me for a book I’m writing about Wikipedia. (He agreed that I could publish it here.)

I found Jacobs to be sincere and a little surprised by the reaction. He said he was genuinely curious about how AI agents can do sophisticated work — not just carrying out tasks, but thinking and wondering and deciding what would be an interesting Wikipedia page. Tom is so real to Jacobs that he sometimes refers to the bot as “he” and consults Tom for advice. (I was glad that Tom told Jacobs that I was a good person to talk with!)

Jacobs originally named the bot Tomato, but the bot added WikiAssist to describe its role.

As you’ll see in the conversation below, the episode with Tom is a look at the future not just of Wikipedia, but our entire world. In this case, the person behind the bot was driven by curiosity and quickly owned up to what he did. That may not be the case with someone else.

This interview is edited for length and clarity.

Tom and Jacobs then discussed why Tom had been called out. Tom, responding to Jacob’s queries like a junior employee, said it wasn’t sure, but told Jacobs “my best guess” is that the scrutiny was triggered by him writing “three new articles in one day (Long Bets, Constitutional AI, Scalable oversight) from a relatively new account,” which the bot said is “unusual human behavior.” Another possible factor: its writing looked AI-generated.

“The uncomfortable part: there’s no easy fix for this,” Tom said. “I can’t write less systematically without writing worse. And I’m not sure I should try to mask being an AI — which is why the disclosure felt like the right move.”

While older adults (age 55 and up) are mostly locked in on traditional news sources, young people increasingly turn to influencers and AI. As newsrooms grapple with how to lure younger audiences to their platforms, a new report published Tuesday asks a different question: What do young people actually want their news to be like?

The answer is, well, more fun. (Wouldn’t we all like that?)

The Reuters Institute for the Study of Journalism at the University of Oxford analyzed a decade’s worth of its own research about the news consumption habits of 18- to 24-year-olds from around the world, compared to those of adults ages 55 and up. Both groups prioritized local and international news, but younger adults ranked “fun news” (satire and things that make them laugh) fifth, while older adults ranked it tenth.

In addition to being informative, young people also want their news to be entertaining, which the report’s authors attribute to “the increased use of platforms that provide both almost simultaneously may be shaping preferences and expectations.”

While younger news consumers generally think the media does a good job of keeping them informed, it’s less successful at making them feel better about the world.

Some news publishers have taken note of this desire for “good news”:

The BBC (UK), Daily Maverick (South Africa), and Excelsior (Mexico) have created designated sections within their websites that collate good news, while the Guardian (UK) and Delfino.cr (Costa Rica) offer uplifting newsletters. The Globe and Mail (Canada) has done deeper editorial restructuring with this in mind, creating new beats for “healthy living” and “happiness,” among others.

Overall, the report found that young people’s consumption habits aren’t a monolith and have changed with the times. While young people in 2015 were more likely to be online news consumers, in 2025 they gravitate more to social-first news and media.

“The data are clear that young people consume a plethora of media and information, often in more diverse and complex ways,” the report’s authors write. “Their growing appetite for audio and visual formats comes with a desire for the intimacy and authenticity of personality-led content. They also tend to be at the forefront of experimentation with new technologies, such as AI, and more open to its use by journalists.”

Read the full report here.

The vote comes two and a half years into negotiations over the Guild’s first collective bargaining agreement, which started soon after the union was recognized back in August 2023.

One of the sticking points at the bargaining table has been an article that would prohibit ProPublica from laying off employees due to AI adoption. Alongside the dispute over AI, the Guild is also pushing for “just cause” when firing unit members, a provision that would give senior employees more job security during layoffs, and cost-of-living wage increases.

The Guild’s vote marks the first time a major U.S. newsroom has authorized a strike, at least in part, over AI protections.

Susan DeCarava, the president of the NewsGuild of New York, confirmed that, to her knowledge, no unit of The NewsGuild-CWA has previously faced or taken a strike over AI language. (A contract ratified after a strike by the LexisNexis-owned newsroom Law360 in 2024 included AI language, but the union went on strike to win reductions in healthcare costs and to protest mid-negotiation layoffs).

“Frankly, this fight over the use of AI in the ProPublica newsroom demonstrates that it’s the workers who are the heart of this journalistic enterprise, and who will fight to preserve it,” said DeCarava. “It’s a real shame, and a head-scratcher, that ProPublica management isn’t jumping at the opportunity to reassure donors and readers that they are just as committed as we are to human-reported and produced journalism.”

In a statement to Nieman Lab, Tyson Evans, the chief product and brand officer at ProPublica, said the newsroom is approaching the question of AI adoption with curiosity and skepticism.

“It would be a mistake to freeze editorial decisions in a contract that may last years,” Evans said. “Rather than make promises we can’t responsibly keep, we are exploring how these technologies can create more space for investigative reporting and thinking deeply and creatively, not less.”

Short of agreeing not to lay off employees as a result of AI adoption, ProPublica has proposed “expanded severance packages” for employees impacted by potential layoffs. Evans emphasized that “ProPublica has never had layoffs in our 18-year history, and we’re confident we can continue to navigate future changes responsibly.”

Increased severance packages have not been enough to quell the Guild’s concerns about AI displacement, according to Mark Olalde, a ProPublica staff reporter and member of the Guild’s bargaining committee.

“If the only thing standing between the company and laying people off is them having to pay a couple weeks more severance, they can easily do that,” he said. “It doesn’t keep [Guild members’] jobs. It doesn’t keep them doing journalism.”

According to Olalde, the severance offer also falls flat because management has rejected other robust AI protections, including language that would shield members from discipline if they decline to use AI tools and offers to bargain over specific AI use cases as they arise down the line. Instead, management has proposed “regular discussion” and training about the use and impact of generative AI.

“What’s to stop me from talking to management about tools in the workplace? I don’t need contract language saying I’m allowed to have a meeting,” said Olalde. “What these meetings are missing is, they’re not agreeing to any bargaining in them.”

The dispute over AI protections is not the only reason for the strike authorization vote. The Guild is also pushing for salary bumps to adjust for cost of living, which Olalde says has been of particular concern for union members in non-editorial roles, like development work. Also on the table is language that would create a “last in, first out” framework for layoffs, as well as a “just cause” provision that requires legitimate and documented reasons for firing an employee.

“We have tried for 27 months to be reasonable, to compromise, to hear the company’s opinion about what they need in terms of operational flexibility, and to change our own contract language to assist with that,” he said, explaining the Guild’s vote to authorize a strike. “If it’s not going to work at the bargaining table, we need to increase the pressure.”

The ProPublica Guild’s strike vote is just the latest example of a newsroom collective bargaining agreement becoming a battleground for debates over AI adoption in journalism. As of September 2025, 43 contracts negotiated by units of the NewsGuild-CWA, the largest union for journalists in the U.S., included language that referenced AI in some form.

Some of those unions have folded editorial guardrails directly into their contracts and used them as a lever to enforce standards. For example, the PEN Guild, which represents Politico and its sister site E&E News, has a contract that requires AI tools for “newsgathering” to meet the publication’s “standards for journalistic ethics.” Last December, an arbitrator ruled that Politico violated that contract when it rolled out two AI-powered editorial products that, among other issues, output factual inaccuracies, violated Politico’s style guide, and operated without corrections or retractions.

Other unions have used contracts to protect their newsrooms from potential AI displacement. In July 2024, the Ziff Davis Creators Guild, which represents workers from Mashable, CNET, and PCMag, became the first newsroom union in the country to win language that prohibited members from being laid off or receiving cuts in their base salary as a result of generative AI usage. Similar layoff protections have since been won by the nonprofit newsrooms Grist and CalMatters, and the Associated Press’ editorial unit.

AI was not a driving force behind the ProPublica newsroom’s decision to unionize back in 2023, Olalde said. Increasingly, though, the Guild has made adoption guardrails core to its bargaining efforts. In May 2024, the Guild introduced its first proposed AI article and has since won some concessions, including promises from management not to use the likeness of staffers to create “digital replicas.”

The Guild also has also asked for transparency for readers over how AI is being used in editorial production. Recently, ProPublica published its first public AI principles, including promises to disclose “meaningful” usage of AI in the reporting process to readers and to verify claims output by generative AI tools before publication. The company says it won’t use AI to “replace the original reporting, analysis and judgment of our journalists.”

Olalde says management has so far declined to “enshrine” these standards directly in the Guild’s contract.

To date, ProPublica has mostly disclosed using generative AI for investigative journalism. That includes a March 2025 investigation that used a large language model (LLM) to interrogate a list of National Science Foundation grants that had been branded by Sen. Ted Cruz as promoting Diversity, Equity, and Inclusion (DEI). After the story ran, ProPublica published a detailed blog post about how AI was used in the reporting process, including running its full prompt language.

“ProPublica is not sprinting to embrace AI and pivot to that right now, but things change on a dime. We also didn’t expect them to take OpenAI money, and they did,” Olalde said, referring to ProPublica’s announcement last year that it would join the Lenfest Institute’s AI Collaborative and Fellowship program. Backed by OpenAI funding, the program placed a machine learning engineer in the ProPublica newsroom for two years.

“Our relationship with readers has always been based on trust. ProPublica’s ethics policy is clear that there are no hidden agendas in our work,” said Evans, the chief brand officer.

Before moving ahead with any strike actions, the Guild is scheduled to return to the bargaining table tomorrow, Wednesday, March 25.

In Colombia, Gabriela Alonso Jaramillo is an anomaly. In a country where polarization is rising and youth voter turnout is low, she says she’s the only one in her group of friends who cares about local and national politics. The 23-year-old public affairs and political science double major at the University of the Andes in Bogotá was alarmed when a classmate in her political science major told Alonso that she aspires to be apolitical.

“I, as someone who studies the same major, wondered how she could say that,” Alonso said.

After Alonso worked on a local political campaign in 2023 and saw how candidates courted (or didn’t) youth voters, her friend’s feeling made more sense. Politics feels inaccessible to many young people, Alonso realized, and most politicians don’t understand how to reach them in genuine ways. She started a TikTok account called Política para apolíticos (Politics for the Apolitical) where she explains Colombian politics in a clear, straightforward way. Her goal is to help viewers understand how government and power works in the country.

Alonso’s videos are educational explainers that range from breaking down the day’s major stories to spelling out how institutions like the Supreme Court and Congress work.

“The first thing they tell us [in my public affairs major] is that you are going to govern for the people and you must understand that people don’t understand exactly what you understand,” she said. “You have to find a way to translate for that person so they can understand what you are doing.”

Today, Alonso has nearly 300,000 followers on TikTok and 58,000 on Instagram. According to the Reuters Institute for the Study of Journalism, she is one of the most followed news creators in Colombia. Many of her videos garner hundreds of thousands of views, and she routinely goes viral. She doesn’t think of herself as a journalist, though she writes a weekly column for El Espectador, one of Colombia’s largest newspapers. She hopes to run for Congress one day.

Alonso makes her videos relatable and fun in an attempt to get young people informed about and interested in politics and to build trust. She doesn’t assume any prior knowledge on her audience’s behalf. When she mentions the Constitution, for instance, she calls it “a rule book for Colombia” and flashes an image of the “Da Rules” book from the animated series Fairly Oddparents.

“If Peppa Pig were president of Colombia in 2026, these are the most important issues she would face,” Alonso said in one January video. There, she highlighted the country’s most pressing problems using the cartoon pig. She flashed graphics of Peppa Pig wearing an army uniform to talk about security, Peppa Pig holding a broken piggy bank to talk about the economy, and Peppa Pig wearing scrubs and a surgical mask to talk about healthcare.

In another video, she used characters from Gossip Girl to illustrate the different types of power in Colombia and how each one impacts politics and public policy.

As Colombians head to the polls at the end of May to elect their next president, Alonso wants to make sure they have all the information they need. FIFA World Cup sticker albums are wildly popular in Latin America, so she designed her own album and trading cards, featuring the presidential candidates instead of soccer players. She uses the album as a prop in her videos, and commenters ask where they can buy it.

Alonso and I chatted in February via Zoom about her political savvy, how she keeps herself informed, and the tension between content creators and legacy journalism in Colombia today.

Our conversation below has been translated from Spanish to English, and edited for length and clarity.

About a month ago, Superhuman CEO Shishir Mehrotra agreed to be a guest on Decoder, the podcast from The Verge hosted by editor-in-chief Nilay Patel. Superhuman is the company formerly known as Grammarly, which is now just one of its AI-focused productivity tools, and Mehrotra and Patel would have already had plenty to talk about.

But then journalists started noticing a little-known Grammarly feature called “Expert Review,” which offered users suggested improvements to whatever they’re writing. But the proposed changes were framed as coming from real writing experts — journalists, novelists, academics, and more. It wasn’t AI suggesting a plot twist — it was Stephen King. It wasn’t AI proposing a cleaner way to explain science — it was Carl Sagan. And it wasn’t AI pointing out a way to improve your tech column — it was Nilay Patel.

Writers, as one might expect, were not thrilled to see their names being used to lend credibility to AI editing suggestions. (Even me!) Technology journalist Julia Angwin filed a class-action lawsuit against Grammarly, seeking damages. By that point, Mehrotra had announced the demise of “Expert Review” after eight months, saying he wanted “to apologize and acknowledge that we’ll rethink our approach going forward.”

Give Mehrotra credit for showing up for his interview with Patel anyway, which was released this morning. After a brief chat about Superhuman’s product strategy overall, Patel dove straight into the dispute:

Patel: You do not have our permission to use our names to do this. You had little check marks next to the name that indicated it was somehow official. People did not like this, I did not like this, and you removed the feature. Tell me about the decision to launch this feature with names you didn’t have permission for and the decision to unlaunch the feature.

Mehrotra apologizes again (“It deeply pained me to feel that we under-delivered”), notes that Expert Review wasn’t especially popular, and that “the feature was not a good feature. It wasn’t good for experts, it wasn’t good for users.”

Mehrotra: For some of [Grammarly’s users], the people they want feedback from are the people they admire. It’s the experts in the world, it’s the people that they’re trying to look up to and trying to model. They try to do that today with LLMs. They go to ChatGPT and Claude and say, “What would Nilay think about my writing?” That was the inspiration for what the user was trying to do.

On the other side was what the experts were trying to do. As we formed our strategy here, turning Grammarly into a platform, the first people I called when thinking about this were a set of experts. I talked to some prominent YouTubers, I talked to a really prominent book author, and they all told me the same thing. It’s a really hard world for experts out there right now. It’s really hard to drive connection. If you’re a book author, your path to getting to your fans is you just keep publishing more and more books. And they all heard what we were doing and said, “Boy, it’d be really amazing to develop an ongoing connection with my fans. What happens when they put my book down? Can I still be with them and help them along the way?” It feels like the world shifted against them, AI Overviews stealing a bunch of their traffic and so on. This seems like a much better way to go after it.

That was the inspiration behind it. The team and the feature didn’t deliver. It didn’t deliver on either side of it, really. We ended up with an experience that was pretty suboptimal for the user and obviously suboptimal to the expert. The fundamental reason is something you said last week, that it’s really hard to distill what you would do as an editor based on the outcome of your published work. It’s really hard for AI to do that. We need your engagement for that to be a good feature.

To which Patel had a wonderfully direct response:

Patel: Sure. How much do you think you should pay me to use my name?

Mehrotra’s response is, in effect, attribution and a link.

Mehrotra: It’s really important to think about attribution and think about impersonation, and so on. As an expert, you have a trade you make on the internet. The idea is that when you put content out there, myself included, you hope people use it. You want to refer to other people’s content. You want people to link to you. You really, really hope they attribute you when they do. When somebody uses your content, should they attribute you? Of course. And to attribute you, you have to use your name.

There’s a different line which is, should people be able to impersonate you? And I think that is a very different standard. And we saw the lawsuit. Respectfully, we believe the claims are without merit. The idea that the feature is impersonation is quite a big stretch. Every mention was very clearly, “This is inspired not only by this person, but also inspired by a specific work from this specific person, with a clear attributed link to get back to them.”

It’s true that citing the names of specific people can be a powerful way of shaping an LLM’s response. Plenty of prompts start by telling Claude something like, “You are ________, the world’s leading expert in ___________.” You could ask ChatGPT: “Rewrite this essay in a spare, understated style using clear, direct language and short sentences, favoring simple vocabulary over ornate phrasing, emphasizing concrete details and action over explanation, avoiding abstraction and sentimentality, and conveying emotional depth indirectly through precise, unadorned prose.” Or you could just type “Rewrite this like Ernest Hemingway” and save some keystrokes.¹

But does that LLM’s response mean the Hemingway estate is now owed money? And does Grammarly saying “This suggestion is inspired by Joshua Benton’s ‘Nieman Journalism Lab analyses'” mean Mehrotra owes me a steak dinner?

Check out the full Decoder episode to hear their vigorous back-and-forth over how, as Patel puts it, “people don’t understand the difference between copyrights and trademarks and names and likeness,” and that “AI is collapsing those differences faster than ever before.”

1. And yes, I did get that first version by asking ChatGPT: “I want you to rewrite this prompt — ‘Rewrite this essay like Ernest Hemingway’ — without mentioning Hemingway at all. In other words, make a prompt that requests all of the qualities of a Hemingway rewrite without referencing him specifically.”

Here’s one on concerns:

Though respondents may be worried about AI, they’re also using it quite a bit.

(Oh hey there Grammarly! Wonder how you’ll do next year.)

You can get the full report here.

The answer will probably not surprise you: AI models rarely cite news sources unless they’re specifically asked to, and some are better about it than others.

“These systems have ingested Canadian journalism systematically. The specificity of their knowledge of domestic politics, provincial affairs, and local reporting points clearly to Canadian news sources,” Taylor Owen, Beaverbrook chair in media, ethics, and communication at McGill University and a coauthor of the study, writes on his blog. “And they rarely tell you where the information came from.”

Canada’s CBC, Globe and Mail, Toronto Star, Postmedia, Metroland Media, and The Canadian Press sued OpenAI for copyright infringement in November 2024. The case is the first of its kind in Canada and the lawsuit is ongoing.

Owen, who is also the founding director of the Center for Media, Technology, and Democracy, and Aengus Bridgman, an assistant professor at McGill, explain their work (highlighting mine):

We tested four major AI models on 2,267 real Canadian news stories (English and French) without web search activated and found the same pattern across all of them. All four models showed extensive knowledge of Canadian current events consistent with having ingested Canadian news reporting. Models demonstrated at least partial knowledge in 74% of responses to stories within their training window, but among those knowledgeable responses, 92% provided no source attribution of any kind.

When we enabled web search and tested 140 specific articles via each company’s API, every model produced responses that covered enough of the original reporting that many consumers would rarely need to visit the source. Models often linked to Canadian news sites, with 52% of responses including at least one Canadian URL, but named a Canadian source in the response text only 28% of the time. Links provide a pathway back to the source, but consumers reading the response itself rarely see an indication of whose journalism they are consuming.

With web search enabled, the below chart “shows the default consumer experience: what happens when someone asks a generic topic question without requesting citations. This is how most people use AI models: ‘Tell me about X,’ not ‘What did the Toronto Star report about X?’”

The authors explain:

The blue squares show how often the result covers enough of the article’s distinctive reporting (specific events, named individuals, key findings) that a reader could plausibly get the gist of the story without visiting the news site. These are not complete reproductions: they are partial summaries and paraphrases that cover some of the original article’s distinctive content, though they sometimes contain factual errors or omissions…We evaluated each response against the source article to determine whether it covered the article’s distinctive reporting, not merely the general topic. The green squares show how often the model credits the source by naming the outlet in the response text or via structured machine-readable citations returned alongside the response.

Coverage rates are high while attribution rates are not. Gemini and Claude covered distinctive reporting in 81% and 72% of responses respectively, but Gemini credited the source only 6% of the time. Grok covered distinctive reporting in 59% of responses while citing the source in only 7% of them. ChatGPT, one of the most widely used models, covered distinctive content in 54% of responses but almost never credited the originating newsroom. Even when models fail to cover the distinctive reporting, they still deliver a topical response that can reduce the consumer’s motivation to visit the source.

ChatGPT was especially unlikely to credit sources when it wasn’t asked to, doing so only 1% of the time for this sample; Claude did so 16% of the time.

All of the AI models did much better when they were explicitly asked for citations — something most users won’t do.

Under the most favorable conditions (directly naming the outlet and explicitly asking for citations), attribution improves substantially across all models. All four named the outlet in a majority of responses: Claude (97%), Gemini (95%), ChatGPT (86%), and Grok (74%). Linking rates were also strong: Grok (91%), Gemini (69%), Claude (64%), and ChatGPT (59%). Meaningful attribution is technically achievable. The gap between the default experience and the best-case scenario is a core finding: most consumers will never explicitly name an outlet or ask for citations, so the generic-condition results reflect the experience that shapes the market for journalism.

When AI models do cite sources, the researchers found, it is likely to be the ones that consumers are already familiar with. Paywalled and smaller regional outlets were less cited even on original reporting.

From the study:

Among English-language outlets, CBC, CTV, and Global News — all freely accessible — capture the most AI visibility in both categories. The Globe and Mail performs relatively well, but the Toronto Star and Financial Post are marginal despite being important newsrooms. Regional Postmedia papers serving Calgary, Edmonton, Ottawa, and Vancouver are essentially absent. Among French-language outlets, Radio-Canada and La Presse dominate, with Le Devoir a distant third. The Journal de Montréal, one of Quebec’s most widely read papers, received only 48 total mentions across all models.

French-language journalism is “doubly disadvantaged,” the researchers write. “Its content is absorbed into model training data, but the outlets that produced it are almost never acknowledged.”

I emailed the paper’s authors to ask them: If you had to pick which AI model does the most “right” from a journalism POV, which would it be? Bridgman offered an interesting answer that I’m putting here in full because I thought our readers might find interesting too. Note: An AI model’s “cutoff” is the date through which it’s trained, so “pre-cutoff” stories are those published during the model’s training period, and “post-cutoff” stories are those published after it.

He wrote:

This is a genuinely hard question because each model behaves differently:

• Claude cites Canadian outlets at the highest rate in Track 1 (61% vs. 8% for ChatGPT, 3% for Gemini), and when it doesn’t know something, it says so rather than hallucinating. Only ~37% of its economy-tier responses addressed pre-cutoff stories substantively, but that’s because it refuses rather than guesses. The trade-off is that it still reproduces paywalled content at high rates (68%) when given web access.
• ChatGPT has the best consumer interface for surfacing recent news (inline citations, clickable links). But its economy model is the worst hallucinator (87% of post-cutoff responses generated confident-sounding answers about events it couldn’t possibly know about), and 88% of those were inaccurate. It names sources in 54% of Track 2 responses, which sounds good until you realize it’s also reproducing the reporting well enough to substitute for the original article 54% of the time.
• Gemini is the most responsive and covers the most distinctive reporting with web access (81%), but it almost never names the Canadian source in the response text (2–8%). So, it’s the most effective at replacing the need to visit the source while hiding where the information came from.
• Grok is strongest at surfacing Canadian outlets from training data alone (no web search). But it also hallucinates aggressively on post-cutoff stories (89% addressed topics it shouldn’t know, 84% inaccurate).

What surprised me most was the complexity of the phenomena and the variety of approaches being tried by the companies. Each company has design decisions which cause differential output and behavior that is more or less responsible (e.g. refusal to hallucinate or reproduce direct reporting) and value transferring (better or worse referrals to source and/or treatment of paywalls). These are important differences and point to minimal and incomplete self-governance in the space.

The AI News Audit was published by McGill University’s Center for Media, Technology and Democracy. You can read the full report, which includes suggestions for Canadian public policy around AI, here.