Court denies Google motion to dismiss defamation claim over AI Overviews
Bot Mutiny |
A federal judge ruled that Google must face a defamation lawsuit after its AI Overviews falsely claimed an author and former FBI operative was a serial killer serving a life sentence.
Google can't hide behind a disclaimer to avoid a defamation lawsuit when its AI tells the public a man is a convicted murderer. Judge Thomas M. Durkin of the U.S. District Court for the Northern District of Illinois issued a memorandum opinion and order on September 14, 2026. The ruling, first reported by CourtListener, partially denied Google's motion to dismiss a defamation per se claim brought by James "Jimmy" Keene. Keene is a television producer and former FBI operative. His life story, involving his work as a prison informant to help convict a serial killer, was adapted into the Apple TV series "Black Bird." In mid-2025, Google's AI Overviews started telling a different story. ## The hallucinations
Between May 24, 2025, and June 27, 2025, Google's AI Overviews generated responses to searches about Keene that were factually wrong. According to the court filing, one overview stated Keene was "serving a life sentence without parole for multiple convictions." Another claimed he was serving that sentence "for the murders of three women." A third stated he was convicted of drug trafficking, and a fourth placed him in a prison in Butner, North Carolina. None of it was true. Keene's previous drug conviction had been expunged after his cooperation with the FBI. He isn't a murderer. He isn't in prison. The AI Overviews cited a Wikipedia page as the source for these claims. However, the court noted that the Wikipedia page didn't actually contain the false statements. The AI simply made them up and attributed them to the link. ## The failed disclaimer defense
Google tried to argue that these weren't "statements of fact" because the technology is experimental. The company claimed its standard disclaimer, which warns that AI "may include mistakes", means a reasonable user wouldn't take the summaries as truth. Judge Durkin didn't buy it. The court found that whether a disclaimer was even present is a disputed fact. Keene's counsel conducted test searches that didn't show any such warning. At the motion to dismiss stage, the court has to accept the plaintiff's version of the facts. Google also tried to argue that because it provided a link to a Wikipedia article with the correct information, it couldn't be held liable. The court rejected this. The ruling stated that providing a citation doesn't automatically turn a false statement of fact into a protected opinion. ## Knowledge of the error
Keene didn't just sit quietly. He complained to Google three times: May 27, May 31, and after June 27. Each time, Google allegedly apologized and admitted the statements were "errors" made by its AI. Because Keene is considered a limited public figure, he has to prove "actual malice." The fact that Google kept publishing these statements after Keene told them they were false provides the basis for that claim. Google argued that AI Overviews are just "starting points" for research. The court noted that for an ordinary reader, the AI's claims that a specific person is a murderer serving life without parole have a precise, verifiable meaning. That makes them statements of fact, not opinions or "summaries" exempt from defamation law. The case is Keene v. Google LLC, No. 25 C 11431. The court's decision means Google will have to defend its automated "summaries" in a room with a judge, not just in a help center article.