A federal court will allow a man to sue Google for defamation after the company's AI Overview falsely told the world he murdered three women and was serving a life sentence — a ruling that carves a crack in the legal shield Big Tech has used for decades to dodge accountability for the damage their own machines do to ordinary Americans.
James Keene — a writer, TV producer, and former FBI operative — discovered last year that Google Search queries about him returned AI-generated responses claiming he had killed three women and was serving a life term in prison, according to the Boston Herald. Keene had indeed served time; his experience behind bars alongside a suspected serial killer became the basis for the Apple TV+ miniseries "Black Bird." But his actual crime was marijuana-related, and he was released after 17 months. The AI didn't just get details wrong — it invented a triple-homicide out of thin air and attached it to a real person's name.
Keene "repeatedly alerted Google that the AI Overviews could be defamatory," court documents say, but the false statements remained accessible for at least two months before he filed suit seeking at least $250 million in damages.
Google moved to dismiss the claim on two fronts. First, the company argued that AI Overviews are not considered statements of fact — a remarkable position for a product that presents itself as an answer engine at the top of the world's most-used search page. Second, Google claimed Keene qualifies as a "limited public figure," which would require him to "plausibly plead" actual malice on Google's part.
The U.S. District Court in Illinois wasn't buying it wholesale. The opinion issued Monday allowed two of Google's motions to dismiss and denied two others, stating that Keene may file an amended complaint and proceed with his case.
The stakes here extend well beyond one man's reputation. For years, the tech industry has hidden behind Section 230 of the Communications Decency Act, which shields platforms from liability for content posted by users. That shield was always controversial — the tavern debates of the founding era assumed that the man who speaks is the man responsible — but there is at least a logical argument that a platform hosting third-party speech shouldn't bear the publisher's liability. AI-generated content blows that logic apart. Google's AI Overview isn't a user comment or a linked article. It is content Google's own system invents, synthesizes, and presents as authoritative fact at the top of every search result. When the machine lies, the machine's owner should answer for it.
Google's defense — that AI outputs aren't statements of fact — reveals the game: the company wants the trust and prominence of an answer engine without the legal responsibility of a publisher. If a newspaper printed that an innocent man was a triple-murderer, it would face ruinous liability. Google's AI does the same thing, at planetary scale, and asks the courts to call it something else.
The court's partial green light doesn't guarantee Keene will win. He still faces the burden of amending his complaint and proving his case. But the ruling establishes that AI-generated defamation is at least legally actionable — and that Big Tech can't simply algorithm its way out of the consequences.
The open question: will courts follow this logic to its conclusion, or will the lobbying class and the think-tank ecosystem find a way to carve out a new immunity before ordinary Americans get their day?
