NewsGuard has identified more than 3,000 AI content farms—more than double what it could find a year ago using manual techniques—and it’s now partnering with AI detection startup Pangram Labs to scale that tracking as the problem accelerates.

AdWeek reports the new detection tool, announced Thursday, uses Pangram’s proprietary models to scan not just individual pages but entire domains for signs of AI-generated content at scale. When Pangram flags a site, NewsGuard analysts review it manually before applying a formal “AI content farm” designation. Sites qualify when a substantial share of content appears AI-generated, there’s no disclosure to readers, and the site’s presentation could convincingly pass as human-produced journalism.

The scale of the problem is striking. Between 300 and 500 new AI content farm sites are emerging every month, according to Pangram. Many operate under generic news-adjacent names (e.g. Times Business News, Business Post) and publish misinformation about real brands, politicians, and public health. In one case, a site called Citizen Watch Report falsely claimed two U.S. senators spent $814,000 on hotels in Ukraine; the story was amplified by Russian state media before being debunked.

Another site falsely claimed Coca-Cola threatened to pull its Super Bowl sponsorship over a halftime show for which Coca-Cola wasn’t even a sponsor. Both sites ran ads from major brands.

That last detail is the commercial mechanism. Most of these sites are made-for-advertising (MFA) operations—cheap content churned out to capture programmatic ad spend. In a two-month period, NewsGuard found 141 blue-chip brands advertising on AI content farm sites. The slop economy runs on their budgets.

“If we can’t detect AI content, then every communication space is going to be flooded with inauthentic content that’s cheap to produce and difficult to impossible to differentiate [from] something authentic,” Max Spero, Pangram’s CEO, told AdWeek’s Kendra Barnett.

NewsGuard’s detection data will be available for advertisers to license directly or through their agencies, with a pre-built integration into The Trade Desk for pre-bid blocking. A consumer-facing browser extension integration is also under consideration. Pangram, founded in 2023 by a former Google engineer and an ex-Tesla scientist, gained independent validation when a Nature report last September found it highly capable of flagging AI-generated academic papers.

The detection arms race is worth watching. Early AI content farms were easy to spot — sites would publish articles containing ChatGPT error messages verbatim. Today’s operations are more sophisticated. The tools to catch them are getting sharper too, but the math still favors the farms: generating slop is cheaper and faster than reviewing it.

The post NewsGuard and Pangram are building an AI slop detector as content farms multiply appeared first on The Media Copilot.

Ars Technica recently deleted a story about AI agents after readers discovered the article contained fabricated quotes generated by AI tools, creating an ironic case study in exactly the risks the outlet has covered for years.

Benj Edwards, Ars Technica’s senior AI reporter, used an experimental Claude Code-based tool and ChatGPT to help extract quotes from a two-page blog post while working sick with COVID and a fever. The AI hallucinated paraphrased versions of quotes rather than providing the source’s actual words.

“The irony of an AI reporter being tripped up by AI hallucination is not lost on me,” Edwards wrote in a statement assuming full responsibility.

The story covered Scott Shambaugh, a coder who claimed an AI agent wrote a hit piece about him after he declined its code contributions. Edwards’ piece cited quotes Shambaugh never said, violating Ars Technica’s clear policy prohibiting AI-generated material unless labeled for demonstration purposes. This is a stark example of an AI agent experiment gone wrong.

Editor-in-chief Ken Fisher called it “a serious failure of our standards” and noted the outlet has “covered the risks of overreliance on AI tools for years.”

The incident highlights several newsroom risks. Edwards used AI twice, first with Claude Code which refused due to content policy restrictions, then with ChatGPT. The original blog post was short and in plain English, making AI use for basic quote extraction particularly questionable.

Ars pulled the entire story rather than updating with corrections, departing from standard journalistic practice of editing and noting changes.

For newsrooms, the lesson is stark: AI tools cannot reliably perform basic journalism tasks like accurately citing sources. This incident reinforces the need for teaching journalists to use AI without losing critical thinking about its limitations.

The fabricated quotes violated both professional ethics and company policy, demonstrating that AI hallucinations remain a fundamental liability even for reporters who cover AI’s limitations daily.

The post Ars Technica pulls story after discovering AI hallucinated quotes appeared first on The Media Copilot.

If you run any kind of media business in 2026, you develop a strange new hobby: speed-running your own gullibility. Every week—honestly, most days—something drops into my inbox offering to “unlock” my growth: more newsletter subscribers, a bigger podcast audience, a fatter pipeline of client leads. I’ve learned to treat these pitches like background noise. Still, a few are so polished they feel tailored, the kind that poke right at the soft spots (“you’re leaving so much on the table,” et al.). I never reply. But I do occasionally catch myself asking the annoying question: Which of these are real?

`A few months ago, I decided to outsource that doubt. I was reading one of those emails and opened the Assistant sidebar in my AI-powered browser. I typed, “this look sus?” The assistant didn’t hesitate. Yes, it said: the pitch—about finding funding for The Media Copilot—left out basic details any legitimate org would include. And the sender? An email address tied to a nonexistent domain, plus no LinkedIn profile. Not subtle. Just efficient.

`That moment stuck with me as I read in Time about a team at MIT that runs an online portal tracking harmful AI incidents. Their running tally makes the trend hard to ignore: the use of AI to cause harm, intentionally or not, has increased significantly over the past few years. Some of it is garden-variety error, some of it is deliberate. The fastest-growing buckets are the ones you’d expect: misinformation and malicious actors. If your goal is to mislead, misinform, or straight-up scam people, it’s never been cheaper—or easier—to operate at scale.

In theory, this is where journalism steps in. After all, one of the media’s jobs is to provide a check on misinformation. When those Biden robocalls were making the rounds, for example, the debunking was swift. But that’s the highlight reel. Most incidents don’t go viral, don’t make national headlines, and don’t trigger an army of fact-checkers. Meanwhile, the number of journalism jobs keeps shrinking, and the reporters who remain have the same constraint as everyone else: finite bandwidth.

Doubt needs direction

As misinformation from AI scales up, it’s creating a world where everyone is increasingly skeptical of what they read, see, and hear. That reflex is understandable—and corrosive. Last year, a paper from the National Bureau of Economic Research found that exposure to AI-driven misinformation led to less trust in media in general. So yes, skepticism is spreading. But skepticism, on its own, doesn’t produce clarity. It produces exhaustion.

This is where the media can still matter, even if it can’t possibly chase every fake. The most valuable move isn’t to debunk every deepfake or scam, which is clearly a losing battle. It’s to teach people how to aim their skepticism. There’s value in having a simple method for stress-testing what you see without spiraling into a “nothing is true” worldview.

The irony is that the verification tools are no longer locked inside newsrooms. They’re sitting in everyone’s browser, in everyone’s phone, baked into the same AI systems that are helping bad actors crank out lies. These tools can quickly check sources, analyze claims, and surface supporting evidence. That doesn’t mean you should treat a chatbot like an oracle about a story. But it does mean AI can be used as a lens, one that nudges you toward better questions, not instant certainty.

Think about my email example: The assistant didn’t “decide” what was true; it did the tedious work fast—looking up subjects, flagging inconsistencies, and pointing me toward new questions. That’s journalism, minus the byline. And if journalists can translate that mindset into practical guidance, readers get something better than a one-off debunk. They get a repeatable habit that helps them spot bad info, and avoid reflexively tossing the good info, too.

Keeping your guard up without giving up

So what does an “AI verification layer” actually look like in the wild? Start here: skepticism is the beginning of the process, not the finish line. Used well, it’s a tool for interrogation. Used poorly, it’s a shortcut to confirmation bias, where every vague suspicion becomes “proof” that you were right to distrust everything. Below are three habits, each rooted in basic journalistic principles, that work with almost any AI tool.

• Ask the same question twice: A lot of AI harm doesn’t begin with malice. It begins with a user asking something ordinary, then getting nudged down a rabbit hole that gets darker or weirder with each turn. Sometimes it ends tragically. One simple way to interrupt that slide is to ask the same question again, but rephrased or reframed. Then compare what you get back. If the answers materially disagree, don’t hand-wave it away—treat the inconsistency as the story.

• Force specificity: Good interviewers don’t let big claims float by unchallenged. When someone declares something sweeping, they press for the who/what/when. Do the same with AI. Ask it to make the claim more specific. What supports that assertion? Who was involved? What are the underlying facts? When did it happen? If the tool can’t move from generalities to concrete details, it’s a signal that the information might be thin, shaky, or invented.

• Spot-check sources: If a claim hinges on something “out there on the internet,” verification shouldn’t be an epic quest. Follow the link. Look for the primary source. Cross-check a key detail. If you can’t confirm it in a minute or two, pause before you share it or build an opinion on top of it. Yes, there are exceptions—anonymous sources exist, and some real information is genuinely hard to verify. But friction is informative. When everything gets slippery, that’s the moment to slow down.

Between AI hallucinations, deliberate disinformation, and the way meme culture blurs seriousness into vibes, it’s no wonder skepticism is becoming the default posture. But without a few guiding principles, skepticism doesn’t stay healthy for long. It curdles into cynicism. Journalists may not be able to verify all the things we want them to. Still, the discipline behind their work—the questions they ask, the standards they lean on—can be taught. And if those habits spread, news consumers can learn to separate good information from bad, even at scale.

The post The worst thing AI did to misinformation was make it ordinary appeared first on The Media Copilot.

BBC News is closing BBC Trending, its 13-year-old investigative unit focused on social media and online disinformation, Deadline reported Monday. Four journalists are losing their jobs.

The timing is hard to ignore. Recently, BBC Trending published an investigation revealing how Facebook grifters were profiting from AI-generated Holocaust imagery. The team also recently examined the phenomenon of Down’s Syndrome deepfakes.

BBC Trending launched in 2013 as what the broadcaster called “the bureau on the internet.” The unit published investigations on the BBC News website and produced a regular radio show for BBC World Service. One former journalist described it as a team that “married artful and considered storytelling with meticulous investigative journalism.”

The closure is part of broader cuts at BBC World Service, which announced plans last year to eliminate 130 jobs to save around £6 million ($8.3 million) by the end of March. BBC News declined to comment on the decision.

BBC Trending was among the few units at a major broadcaster dedicated to tracking viral disinformation and platform manipulation. As AI-generated content floods social platforms, newsrooms are losing one of the few organizations that actually tracked it. Who will fill that gap? 

The cuts come as AI-generated images and deepfakes become harder to detect and easier to produce. Verification work is getting more difficult precisely as the teams doing it are being downsized.

The post BBC shuts down social media investigations unit as AI disinformation surges appeared first on The Media Copilot.

Fake AI-generated videos of Venezuelans celebrating President Nicolás Maduro’s capture flooded social media this week. The clips looked real enough. But the more troubling part wasn’t how convincing they were.

It was who was sharing them.

As John Herrman writes in New York Magazine’s Intelligencer, Elon Musk widely shared and reposted the fabricated celebration videos to his massive following. That’s notable because Musk is “perhaps the best equipped in the entire world to know that they weren’t real.” He runs an AI company connected to a social network with a fact-checking system.

The old hypothesis about AI disinformation assumed bad actors would trick unsuspecting masses with realistic fake videos. The reality is stranger.

“They were passed around by a powerful person with a large following…for an audience of people who do not care whether they are” real, Herrman writes.

Even X’s own chatbot Grok got duped. Reuters reports the bot initially told users the video “shows real footage of celebrations in Venezuela.” It later acknowledged the video “appears to be manipulated” only after being challenged.

The clips served as what Herrman calls “ambient ideological slop for backfilling a desired political reality.” They weren’t meant to persuade skeptics. They gave true believers something satisfying to share.

This creates a new challenge for newsrooms. Traditional fact-checking assumes people want the truth but lack access to it. What happens when the audience treats reality as optional?

The answer may be sobering. AI disinformation isn’t primarily causing chaos or confusion. Instead, Herrman writes, it’s creating “a perverse form of order in which persuasion is unimportant, disinformation is primarily directed at ideological allies, and everyone gets to see exactly what they want.”

The post AI-generated Venezuela videos reveal a darker truth about disinformation appeared first on The Media Copilot.

Within hours of Nicolás Maduro’s arrest by U.S. forces, fake AI-generated images of Venezuela’s ousted president spread across social media faster than newsrooms could verify them.

The incident marks one of the first times synthetic imagery has depicted a major figure during a rapidly unfolding news event, according to a New York Times report by Stuart A. Thompson and Tiffany Hsu.

“This was the first time I’d personally seen so many A.I.-generated images of what was supposed to be a real moment in time,” Roberta Braga, executive director of the Digital Democracy Institute of the Americas, told the Times.

Some fake images made it into Latin American news outlets before being quietly replaced with an official photo shared by President Trump. NewsGuard, which monitors the reliability of online information, tracked five fabricated images and two misrepresented videos that collectively drew more than 14.1 million views on X in under two days.

The Times tested a dozen AI generators and found most tools, including Google’s Gemini, OpenAI’s ChatGPT, and X’s Grok, quickly created fake arrest images despite stated policies against misleading content.

Jeanfreddy Gutiérrez, who runs a fact-checking operation covering Venezuela, said the fakes spread through “almost every Facebook and WhatsApp contact” he has before official images were available.

“It just took a lot of work, because we always lose the battle to convince people of the truth,” Gutiérrez told the Times.

A pattern of failures

That battle is getting harder. The Maduro deepfakes emerged just days after Grok became the center of a global regulatory firestorm. Governments in the EU, UK, France, India, Malaysia, and Australia have all launched investigations after the chatbot began generating non-consensual sexualized images of women and minors at scale.

Bloomberg reported that Grok was generating thousands of “undressed” images per hour earlier this week. The official Grok account posted an apology on X, writing that it “deeply regret[s]” generating sexualized images of girls “estimated ages 12-16.”

If a major image generator can’t prevent the creation of child sexual abuse material, its safeguards against political deepfakes are likely just as porous.

Gutiérrez said many people refused to believe the official image of Maduro posted by Trump was real.

“It’s funny, but very common,” he told the Times. “Doubt the truth and believe the lie.”

The post Journalists scramble as AI-generated images beat official photos to publication appeared first on The Media Copilot.

UK regulators are scrambling to contain a deepfake disaster unfolding on Elon Musk’s X platform. Ofcom, the UK’s independent regulator for communications services, made “urgent contact” with xAI this week after reports that Grok, the platform’s AI chatbot, has been generating explicit images of women and children without consent.

The tool reportedly put images of Princess Catherine, celebrities and ordinary women into sexualized contexts. Users discovered they could digitally undress anyone by simply tagging Grok in a post.

Journalist Samantha Smith told the BBC she felt “dehumanised and reduced into a sexual stereotype” after discovering Grok users had targeted her photos. When she posted about the experience, others asked Grok to generate more.

Technology Secretary Liz Kendall called the situation “absolutely appalling” and called on xAI to to “urgently deal” with its chatbot. She has also backed Ofcom to take enforcement action. The European Commission labeled the outputs “illegal” and “disgusting.”

xAI’s response to journalists has been an auto-reply. Many media organizations seeking comment from the company reported receiving auto replies that said only, “Legacy Media Lies.” But under pressure from regulators in the UK, EU, India, France and Malaysia, the company has issued reactive statements. X’s Safety account said it removes illegal content and works with law enforcement. Musk posted that users who prompt illegal content will face consequences. The Washington Post reported he also responded to one complaint with a laughing emoji.

Call it “breaking verification” instead of breaking news.

This crisis illustrates why the Reuters Institute’s 2026 predictions matter for newsrooms. Harvard Shorenstein Fellow Shuwei Fang told the Institute that news organizations will discover their next product isn’t content but process: answering “Is this real?” at speed.

When fake images spread instantly and AI tools generate convincing forgeries on command, audiences need trusted sources who can quickly establish what’s authentic. News organizations with verification expertise have a product the market desperately needs.

Law professor Clare McGlynn of Durham University told the BBC that X “could prevent these forms of abuse if they wanted to” but “appear to enjoy impunity.”

In a landscape where platform owners dismiss press inquiries as lies, journalism’s verification function becomes essential infrastructure.

The post Grok’s deepfake crisis shows why 2026 is the year of ‘breaking verification’ appeared first on The Media Copilot.

The Associated Press rolled out AP Verify on Monday, packaging AI-powered verification features alongside traditional authentication tools in a single web-based dashboard.

The platform combines reverse image search, frame-by-frame video analysis, and social listening with AI-driven geolocation, object detection, transcription, and a chatbot assistant. It also includes generative AI text detection and team collaboration features.

“In an era of rampant misinformation and digitally altered content, verification is more essential than ever,” said Gianluca D’Aniello, AP’s senior vice president and chief technology officer.

“AP Verify equips journalists with the essential tools they need to assess online content quickly and accurately all in one place – whether it’s identifying the source of a photo, analyzing video or vetting text.”

AP has used the tool internally for a year before offering it to other publishers. The newsroom used it to secure original Texas flood footage by tracing it to its source, verify a viral meteor sighting in South Carolina, and find eyewitness video from the Charlie Kirk assassination, according to Aimee Rinehart, senior product manager for AI strategy at AP.

Before AP Verify, journalists relied on a “patchwork of tools” including Google reverse image search, Rinehart told Press Gazette. The platform integrates third-party providers including Google’s Fact Check, Trint for transcription, Graylark’s geolocation, GPTZero’s AI text detection, Trendolizer, and identity solutions provider Pipl.

“None of the tools are 100 percent,” Rinehart said. “We would never recommend you go straight to publish just based on that tool’s information.”

The platform surfaced a need among under-resourced newsrooms. During market research, a local broadcaster told AP they sometimes run the wrong tornado video and apologize the next day.

Each newsroom subscription keeps content private. Competitors cannot see what others are verifying.

The launch will test whether publishers want centralized verification tools, Rinehart said. She pointed to last year’s Kate Middleton photo manipulation incident, when five agencies including AP pulled a palace-issued image, as evidence that “AP became a source of truth” for publishers.

“That’s what we really want to get to,” Rinehart said, “is everybody trying to discern what is real, what has been retouched, and what can we trust online?

The post AP launches verification dashboard combining AI and traditional tools appeared first on The Media Copilot.