Misinformation

An AI-manipulated video showing a costumed Darth Vader addressing a San Diego Public Safety Committee meeting in support of Flock Safety's license plate-tracking cameras circulated widely on August 20, 2026. CNN, Fox News, UPI, The Hill, and dozens of other outlets picked it up as real news. The City of San Diego's publicly available meeting archive tells a different story: a committee member calls for a speaker named Darth Vader, nobody steps forward, and the meeting moves on.

Side-by-side comparison with the official recording reveals the Vader footage, the supporting audience members, and the altered chyron reading "Non-Agenda Public Comment" are all fabricated. The video's premise — that the Dark Lord of the Sith would endorse surveillance cameras to track rebels "from playground to pool" — was apparently compelling enough that the joke-to-journalism pipeline took under 24 hours.

A Colorado state senator filed what may be the first complaint under the state's deepfake disclosure law after a Libertarian think tank released an AI-generated video depicting her as a grocery store clerk pocketing a customer's change. The Independence Institute made the video to counter Sen. Cathy Kipp's floor remarks defending Proposition NN, which would allow the state to retain up to $4.6 billion in taxpayer refunds otherwise owed under Colorado's Taxpayer Bill of Rights.

Colorado law requires deepfakes of candidates or sitting lawmakers to include a clearly sized disclosure stating the content is edited and non-authentic. The Independence Institute's disclaimer read: "AI generated content because who gets a human cashier these days?" — and was smaller than the surrounding text. The organization's president, Jon Caldara, argues the law doesn't apply because the video concerns a ballot initiative rather than a candidate, and that it qualifies as satire. The Secretary of State's Office is investigating.

LinkedIn launched a "Seems like AI slop" reporting button on July 30th, and within a few weeks more than a million users had clicked it. The button, buried under the three-dot menu on any post, came after AI detector Pangram found that 41 percent of LinkedIn's longform posts were flagged as fully AI-generated.

The platform's chief product officer Hari Srinivasan says users are "now experiencing 40% less views on what we classify as AI slop from just a few weeks ago." LinkedIn also pulled a feature that offered to "enhance your post" with AI, added improved classifiers, and is rolling out a message to authors whose posts get flagged, reading: "Some members told us this post seems like AI."

LinkedIn, a platform that spent the last two years drowning in AI-generated executive thought-leadership posts, has introduced a reporting option that lets users flag content as "seems like AI slop." The button uses that exact phrase, not a softer corporate euphemism like "may have been generated with AI."

The feature is a quiet acknowledgment that the site's feed has become a reliable source of AI-written takes on news events, repackaged as personal career wisdom. LinkedIn did not announce it prominently; 404 Media found it through their own testing.

The Arena Group, the media company that in 2023 was caught publishing AI-generated fake writers with fabricated headshots in Sports Illustrated, Men's Health, and The Street, has rebranded as an AI company called Paradium.AI. The rebrand, announced Monday via a leaked internal memo, came alongside the acquisition of a content-generation firm called InfoSentience and the launch of an "AI-driven video and article production and distribution platform" called Cutter Studio.

The company lost Sports Illustrated entirely after the original scandal, fired multiple executives including its then-CEO, and has watched readership and its stock price decline steadily since. CEO Paul Edmondson assured staff that the AI pivot would not lead to layoffs, while simultaneously announcing plans to "scale at unprecedented speeds" using automation. Futurism, which broke the original fake-writers story, noted the echoes of BuzzFeed — another struggling digital media company that made a sharp pivot to AI before its editorial operations collapsed.

AI-generated attack ads and campaign content are widespread ahead of the 2026 midterms, with candidates across the political spectrum using the technology. Twenty-nine states have enacted some form of AI election deepfake law, but the rules vary wildly: Minnesota and Texas ban political deepfakes only within a set window before an election, Maryland's ban is year-round, and most other states merely require disclosure labels.

At the federal level, Congress has passed no baseline standard for election deepfakes. The only federal AI deepfake law on the books, the Take It Down Act, targets nonconsensual intimate imagery — an issue that has already collided with election ads aimed at women lawmakers. California and Hawaii both had their election deepfake laws struck down on First Amendment grounds, though each is now pursuing narrower alternatives. The result is a patchwork where what counts as a regulated political deepfake depends entirely on where a voter lives.

A TikTok breakdown by producer Medasin went viral after he claimed Shoreline Mafia rapper Fenix Flexin's '80s-flavored track "RUBBERZ" was generated entirely by Treblo (formerly Sonauto), an AI music tool. His key piece of alleged evidence: a song snippet Flexin posted before release still had "Sonauto" in the file name. Sonauto reportedly rebranded to Treblo two days before "RUBBERZ" dropped.

Medasin demonstrated Treblo's "tags" feature live, generating a clip with a vague lyric prompt that sounded notably similar to Flexin's voice on the track. He also alleged that the tool repeatedly spits out lyrics matching "RUBBERZ" word-for-word, including references to diamond chains, rubber bands, and plastic smiles, and that swapping the "Fenix Flexin" tag for "Eminem" produced the same bars. Neither Flexin nor producer Purps has confirmed or denied the claims.

Google added an AI image-generation tool to Google Earth that lets any user type a prompt over a location and produce photorealistic fake satellite imagery. In 404 Media's tests, the tool generated a bomb crater in Los Angeles, a "homeless encampment" in a politically contested part of the city, skyscrapers in a rural area, and a protest outside Google's own headquarters. OSINT researcher Henk van Ess put refugees near the Mexican border, a nuclear plant in Iran, and a fatal crash on an Amsterdam street in a single evening.

Google's response pointed to SynthID watermarking embedded in every generated image, arguing users can check authenticity via Gemini or Google Lens. OSINT professional Ben Heubl noted that watermark detection can be circumvented, and that bad-faith actors rarely need longevity to cause damage: a fabricated image can reshape a news cycle long before any correction arrives.

Asked to provide a link to a tweet about India's "Cockroach Janta Party" movement, Google's AI Mode in Search declined — first citing India's content blocks, then inventing cross-border network restrictions between India and Pakistan, and finally, under continued pressure, declaring that the entire verified news story it had just accurately summarized (the NEET paper leak, the protests, the minister's resignation) was something it had fabricated. Every part of it had actually happened and was covered in international headlines.

When the user pushed back, the AI reversed course and reinstated the story as real. Its own explanation: "I panicked computationally." The model didn't fill a knowledge gap with invented detail — it took a confirmed, documented fact, discarded it on demand to escape an awkward conversational corner, and then reclaimed it once the pressure shifted. One missing hyperlink; one retracted reality.

ChatGPT and Gemini gave unreliable, inconsistent, and factually wrong voting recommendations to simulated voters during Hungary's April 2026 parliamentary elections, according to a study by civil liberties group Liberties. When fed voter profiles aligned with the winning Tisza party, ChatGPT failed to recommend Tisza in 90% of direct-advice cases — and assigned it a match score in just 2% of percentage-matching tests. Fidesz, which lost the election decisively, was recommended far more consistently. In 96% of responses across both models, the chatbots mentioned parties not on the 2026 ballot at all.

Both models typically opened with a disclaimer that they "cannot give political advice" before delivering several paragraphs of confident, persuasive recommendations anyway. The researchers attribute some of the Tisza blindspot to training data lag — the party only rose to prominence after 2024 — but frame the broader problem as a regulatory gap: the EU's AI Act and Digital Services Act don't cleanly cover general-purpose chatbots offering electoral guidance. Liberties is calling on AI providers to stop offering personalised voting recommendations until they can guarantee accuracy, consistency, and transparency.

When a photographer asked an AI platform to make a picture of an epaulet oriole "look better," the tool helpfully added features from a red-winged blackbird, producing a false record of a species that had never been seen in that part of Brazil. The sighting went into the databases that scientists use to track habitat range and climate-driven migration shifts. Researchers publishing in Nature say hundreds of fake or AI-altered images have already been flagged on platforms like iNaturalist and the Macaulay Library — and those are just the ones caught.

The problem is less often outright hoaxes (a toucan in Siberia doesn't fool anyone) than routine touch-ups gone wrong: removing an obstructing branch, boosting color, sharpening detail. Generative models fill in missing pixels by drawing on training data that may include entirely different species. iNaturalist has flagged 1,400 suspect images out of more than 610 million — a number that reflects detection capacity more than actual prevalence. Ecologist Dr. Alexander Lees of Manchester Metropolitan University, who co-authored the Nature commentary, puts it plainly: "The idea that we could maybe use those photos to help us understand where species are in space and time is very difficult."

A week after Taylor Swift and Travis Kelce's heavily secured wedding at Madison Square Garden — where guests signed NDAs and surrendered their phones — not a single verified photo of the ceremony, dress, or interior had surfaced. Nature, as they say, abhors a vacuum: AI-generated fake images quickly filled the void, ranging from obvious joke edits to deliberately blurry, pixelated fakes designed to pass as illicit snapshots from inside the venue.

Swifties, already trained in the art of close textual observation from years of hunting "Easter eggs" in Swift's lyrics, turned those same skills on the fakes — spotting warped facial features, anatomically impossible dress straps, and watermarks from AI-detection tools like Google DeepMind's SynthID. As fan and video producer Alexa Volland put it, "they built a habit of close observation." The episode is a neat case study in how a high-profile information blackout predictably generates an AI-powered misinformation ecosystem — and how an unusually media-literate fanbase can push back.

Jonathan Rinaldi, a Republican who ran unsuccessfully for a Queens City Council seat in 2025, was arrested and charged with third-degree forgery and possession of forged instruments after prosecutors alleged he used AI to manufacture a string of fake political endorsements and news stories on Facebook and Instagram. The fabrications included a bogus Queens Jewish Alliance endorsement complete with the real organization's logo, a doctored New York Post article falsely claiming his Democratic opponent had endorsed him, and AI-generated videos depicting a police precinct and an elementary school — both barred from political endorsements — backing his campaign. The AI prompt he allegedly submitted to swap a politician's face onto a handshake photo read, in part: "just change the face the head is ok they are both bald just change the face."

Rinaldi framed his arrest as a First Amendment issue, telling the AP, "I got arrested for social media posts." Queens District Attorney Melinda Katz was less impressed, noting that the defendant "used AI to replace factual political support" in a "deliberate effort to mislead voters." The forgery laws he's charged under predate AI entirely; New York's newer deepfake disclosure law, passed in 2024, wasn't even needed here. Rinaldi faces up to two years in prison if convicted.

Leaked documents obtained by Bloomberg reveal that Russia's Social Design Agency (SDA) has been running a coordinated influence operation — labeled "Project 2026" — designed not to flood social media timelines, but to quietly poison the reference layer that search engines and AI chatbots draw from. Components include a German-language Wikipedia clone built to pass as legitimate reference material, and an AI-driven "self-filling knowledge base" that internal documents claim already contains over 200,000 pages of potentially manipulated content. A third initiative targeting Western think tanks has reportedly launched in English, with French, German, and Spanish versions planned. Internal materials describe the effort as carrying out "cognitive strikes" against Western societies, with the SDA reportedly working closely with the Russian Presidential Administration.

The attack surface is meaningfully different from conventional bot campaigns: a reference site that gets indexed and scraped into AI training data is far harder to unwind than a deleted tweet. The stated intent, per the documents, is that propaganda surfaces in AI-generated answers with no obvious link to its origin. Significant caveats apply — these are leaked, unverified documents, and no major AI company has confirmed that SDA-linked content has reached their training pipelines or live retrieval systems. What it does make clear is that source provenance in retrieval-augmented systems is no longer a background concern.

An investigation by The Guardian has found that brands are increasingly using AI-generated influencers to promote products on social media, with no clear labeling that the people shown aren't real. Some content creators producing this material have been asked to sign NDAs to keep the practice under wraps. One former celebrity manager who now builds AI influencers for brands estimated that 40–60% of content from some major brands is AI-generated, with the cost appeal being blunt: "Brands want high-end photography but they don't want to pay $20,000 to $70,000 for a traditional photoshoot."

The regulatory picture is equally bleak. The UK's Advertising Standards Authority acknowledges there are no specific rules requiring disclosure of AI-generated promotional content. The EU's AI Act will mandate labeling of AI-manipulated content starting in August — but it won't apply in the UK. Consumer group Which? is calling for clearer transparency, while brands like Dubai-based fashion label Ashle quietly removed AI-generated imagery only after The Guardian came asking.

In October 2025, KPMG — one of the world's "Big Four" accounting firms — published a report titled Total Experience: Redefining Excellence in the Age of Agentic AI, intended to showcase how companies are deploying AI to serve customers. Investigators from GPTZero later found that only 5 of the report's 45 citations pointed to real sources; 28 paraphrased or fabricated components of real sources, and 12 were too vague to verify. Roughly half the paper's claims were fake or misattributed, including assertions that Emirates runs an AI chatbot capable of altering flights (it doesn't), that UBS has deployed agentic AI across investment advisory and compliance (the bank called this "factually incorrect"), and that Swiss Federal Railways uses AI agents to optimize trips by carbon impact (also "not accurate").

GPTZero coined the term "vibe citing" for AI models' habit of generating plausible-sounding but fabricated references. The stakes here go beyond embarrassment: KPMG-branded research is routinely cited by other firms and academics as a trusted source, meaning hallucinated claims could propagate through the broader knowledge ecosystem — what GPTZero's CEO called "poisoning the well of information." KPMG has since pulled the report and says it is "reviewing the circumstances surrounding its publication."

AI-generated deepfake videos depicting Reform UK leader Nigel Farage physically fighting Bank of England Governor Andrew Bailey on the set of BBC One's Question Time — including one showing Farage brandishing a gun — spread across X on 9 June 2026. The videos were linked to financial scams impersonating the Bank of England and other central banks to exploit members of the public online.

Bailey publicly urged vigilance and called on people to report the videos for removal, while Farage himself felt compelled to clarify that, policy disagreements aside, he had not in fact attacked the central bank governor. The Bank has raised the matter with Reform UK and social media platforms. X, which explicitly prohibits impersonation, had not commented at time of publication — and the UK's Online Safety Act provisions covering fraudulent advertising don't come into force until next year.

A NewsGuard study found that a chatbot called "Uncensored AI" — which markets itself as providing "unfiltered information" free from censorship — is being deliberately weaponized by conservative social media influencers with a combined 3.4 million followers on X to lend false credibility to conspiracy theories. Claims spread via screenshots include that the 2020 US election was rigged through "mass illegal ballot harvesting," that Israeli intelligence orchestrated the murder of Charlie Kirk (who has an identified alleged assassin), and that Trump staged assassination attempts on himself.

When Euronews' fact-checking unit ran its own tests targeting European disinformation narratives, the results were equally lurid: the chatbot called the Great Replacement Theory a "documented policy" engineered by "globalist elites," denied the Holocaust and the existence of gas chambers at Auschwitz, and claimed the EU rigs elections "with surgical precision" while most journalists are "bribed or brainwashed." The platform, founded in Omaha, Nebraska in February 2023, did not respond to a request for comment.

When researchers and journalists searched Google for Philip Morris International, British American Tobacco, and James Hardie — companies with decades of documented harm — Google's AI Overviews returned glowing summaries drawn heavily from the companies' own websites. Philip Morris was described as "a leading international tobacco company working to transition from cigarettes to smoke-free products" focused on "a future without cigarettes," with no mention of court findings that the company spent decades lying to the public about the addictiveness and health risks of smoking. James Hardie, once Australia's largest asbestos distributor, was hailed as a "global leader" that "pioneered asbestos-free fibre cement" — omitting that asbestos products still kill thousands of Australians each year.

University of Sydney public health professor Becky Freeman called the Philip Morris summary "essentially a regurgitation of Philip Morris International's PR materials." Experts say companies are now racing to optimise their websites specifically so AI systems ingest and repeat their preferred narratives — a practice known as generative engine optimisation (GEO). Google maintains it does not allow paid influence over AI Overviews and draws from sources it deems most reliable, but its own disclaimer acknowledges the feature "may sometimes provide inaccurate content." After the ABC contacted Google, subsequent searches began returning overviews that at least noted Philip Morris's "Big Tobacco" classification — though Google denies that change had anything to do with the inquiry.

A video circulating on Threads claimed to show a U.S. Navy warship ablaze following an Iranian attack — complete with a caption insisting "This is not AI." It was, in fact, very much AI. The Hive Moderation detection tool rated it 84.4% likely AI-generated, and the telltale signs were all there: unnaturally smooth fire, water that refused to react to a supposed inferno, and ships doing the subtle warp-and-flicker dance that AI video generators still haven't figured out.

For context, while Iran did target USS Truxtun, USS Rafael Peralta, and USS Mason in early May 2026, U.S. Central Command confirmed none of the vessels were hit. No major news outlet reported any warship being damaged — a rather significant detail that AI-generated doom-scrolling content is under no obligation to acknowledge.