Federal Court Allows Libel-by-AI Claim (Brought by Protagonist of Apple TV’s Black Bird Series) to Go Forward

From today’s decision by Judge Thomas Durkin (N.D. Ill.) in Keene v. Google LLC:

Keene is an author, television producer, and former FBI operative. In 1997, he pled guilty to conspiracy to distribute drugs and was sentenced to ten years in prison. In exchange for a pardon, Keene agreed to work with the FBI as a prison informant to help gather evidence against suspected serial killer Larry Hall. Hall was ultimately convicted, and Keene’s conviction was expunged. Keene published a memoir recounting this experience that was subsequently adapted into an Apple TV miniseries titled “Black Bird.”

Keene alleged that, from May 24 to June 27, 2025, at least four Google “AI Overviews” said false things about him:

1. Keene was “serving a life sentence without parole for multiple convictions.”
2. He was “serving a life sentence without parole for the murders of three women.”
3. He “was convicted of drug trafficking.”
4. He was “serving a life sentence without parole in Butner, North Carolina.”

The first three overviews were allegedly responses to a search for “Jimmy Keene net worth,” and the fourth to a search for “Is Jimmy Keene still alive.”

Each AI Overview cited a Wikipedia page as the source of the information and provided a link to that page. However, the Wikipedia page did not contain the false statements in the AI Overviews. AI Overviews 1 and 2 were seen by various family members, friends, and acquaintances.

Keene complained to Google about the false statements on three occasions: May 27, 2025, May 31, 2025, and after June 27, 2025. Google allegedly apologized to Keene after each complaint and deemed the statements errors made by its Artificial Intelligence.

Keene sued, and the court allowed the case to go forward as to AI Overviews 1, 2, and 4 (with one minor procedural twist as to AI Overview 4 that I’ll mention below). The court held that Keene had sufficiently alleged that an ordinary reader would have understood the statements as factual assertions:

First, Google argues that its standard disclaimer indicates that AI Overviews are merely “starting points” using rapidly evolving technology and not statements of fact. According to Google, a disclaimer appears along with all AI Overviews warning that AI Overviews “may include mistakes.” However, the disclaimer Google references is not identified in the Complaint nor included in the attached exhibits. Indeed, Keene maintains that additional “test” Google searches conducted by his counsel did not include any such disclaimer.

Whether or not a disclaimer was present is a factual dispute not to be determined at this stage. In ruling on a motion to dismiss, the Court must accept all well-pleaded facts as true and draw all reasonable inferences in Keene’s favor. Therefore, the Court’s analysis proceeds without consideration of an alleged disclaimer.

Second, Google argues that users could easily verify that the AI Overviews contain false information based on the linked Wikipedia article or other search results. But Google’s cited cases do not support that merely linking to a source with correct information vitiates defamation….

Google also cites Walters v. Openai, L.L.C. (Ga. Super. 2025), in which a journalist asked ChatGPT, an AI large language model (“LLM”), to summarize a legal document. The journalist intentionally sought out an LLM, had prior experiences when the LLM provided him “flat-out fictional responses,” assented to an acknowledgement that responses may include “incorrect Output,” reviewed multiple disclaimers regarding the potential for information that could be “misleading” or “inaccurate,” and received responses stating that the LLM could not provide the desired information. The court granted summary judgment in favor of the defendant in part because a reasonable reader in the journalist’s position “could not have concluded that the challenged ChatGPT output communicated ‘actual facts.'”

Google’s reliance on Walters is unpersuasive because the context in the instant case is wholly distinct. From the perspective of an ordinary user, asking an AI LLM to generate work product and receiving numerous indicia of falsity is different from a basic Google search.

In Google’s own words, a Google search is meant to “connect a user most efficiently to relevant and reliable information.” Although Google has now developed its own built-in AI LLM to “enhance” a search by summarizing information, an ordinary user searching “Jimmy Keene net worth” on Google is seeking factual information, not AI work product. Only after a user read the linked Wikipedia article would any contradiction become apparent, but without reason to question the AI Overviews, many ordinary users would not conduct that further research.

Google asserts that AI Overviews are just the “starting point” for users to “dig deeper.” To the contrary, factual statements with citations to outside sources provide contextual support that the AI Overviews are summarizing objectively verifiable facts contained therein and additional research is not required.

“If it is plain that the speaker is expressing a subjective view, an interpretation, a theory, conjecture, or surmise, rather than claiming to be in possession of objectively verifiable facts, the statement is not actionable.” But that is not the case here. By pointing out that the statements in the AI Overviews are objectively verifiable, Google concedes the very point it is arguing against—the statements are facts….

Keene didn’t dispute that he was “a limited public figure regarding commentary on his own life’s story,” which meant that he had to adequately allege “actual malice,” which is to say knowing or reckless disregard of falsity. But the court said this requirement was satisfied:

Probative evidence of recklessness includes a publisher’s knowledge of serious factual inconsistencies, as well as the failure to investigate or independently verify disputed or questionable factual assertions.

Keene contends that he has sufficiently pled actual malice because he directly notified Google on three occasions that it was publishing false information about him. Despite this notice, Google continued to publish similar defamatory statements through at least late June 2025. Keene further alleges that Google repeatedly acknowledged and apologized for the false AI Overviews appearing on its platform, thereby admitting awareness of their existence, yet continued to permit their publication with actual knowledge of their falsity or at minimum reckless disregard for their truth.

Google argues that Keene has not adequately pled actual malice because Keene’s allegations are conclusory and he does not allege that he informed the individuals at Google responsible for publishing the AI Overviews about their alleged falsity. See N.Y. Times Co. v. Sullivan (1964) (stating that a plaintiff must “br[ing] home” the state of mind required for actual malice “to the persons in the … organization having responsibility for the publication”). But Keene pleads that he corresponded with Google multiple times regarding the false statements contained in the AI Overviews. Prior to discovery, Keene may not know the names of the responding individuals. But at this stage, allegations of correspondence plausibly indicate that such emails were circulated to individuals at Google responsible for policing the AI Overviews, who then, with knowledge of falsity of the AI Overviews, took no corrective action and the statements continued to be published.

Google also attempts to introduce emails to discredit claims that Google acknowledged and apologized to Keene. However, the introduction of such evidence is not pertinent at the pleading stage. Even if the Court considered the emails, while they do not support the allegations of repeated apologies, they do support that Keene repeatedly alerted Google that the AI Overviews could be defamatory.

Google’s Exhibit 3 shows that Keene filled out and submitted a form to “report a legal removal issue” to Google on May 25, 2025, which described in detail his complaints concerning the AI Overviews. Either this report made its way to an individual at Google, which could plausibly support actual knowledge of falsity, or no human at Google reviewed the legal removal request and subsequent correspondence, which could plausibly support reckless disregard for falsity. Without more information, the emails are not dispositive. At this stage, the Court need only determine whether Keene has plausibly alleged actual malice. He has.

The court noted that Keene had failed to “plausibly allege publication to a third party” as to AI Overview 4, which Keene said was a “mere clerical oversi[ght]” in drafting the complaint. Because of this, the court dismissed the claim as to Overview 4, but allowed Keene to amend the complaint to resolve that problem.

The court held, though, that AI Overview 3 was substantially true and therefore wasn’t defamatory:

AI Overview 3 states that Keene “was convicted of drug trafficking.” In fact, Keene was convicted of felony conspiracy to distribute cocaine…. “[D]rug trafficking” is a term in common usage that encompasses drug distribution often as part of a larger enterprise…. [And t]he fact that Keene’s conviction was later expunged does not negate substantial veracity.

For more on other such recent lawsuits, see my Large Libel Models posts; for more on the legal issue, see Large Libel Models? Liability for AI Output.

Jeffrey B. Steinback and T. Paul S Chawla represent Keene.