Google faces court battle over AI hallucinations
Conservative activist Robby Starbuck says Google's AI falsely accused him of crimes, including rape. A judge has rejected Google's bid to dismiss the lawsuit.
Image by Shutterstock.
- A Delaware court allowed Robby Starbuck’s defamation lawsuit against Google over AI-generated statements to proceed.
- Starbuck alleges Bard, Gemini, and Gemma generated false claims about sexual assault, Epstein ties, and criminal activity.
- Google says the claims involved AI hallucinations, some addressed in 2023, and denies replicating them in consumer products.
- The case could shape when AI companies face liability for defamatory content created by their models.
Key Takeaways by nexos.ai, reviewed by Cybernews staff.
Google will have to defend a lawsuit brought by Robby Starbuck over allegedly defamatory AI-generated statements, a Delaware court ruled on Friday.
Last October, conservative activist Starbuck filed a lawsuit against Google over what he described as "radioactive lies" generated by its AI products, including Bard, Gemini, and Gemma.
"When users submit queries to Google’s AI platforms about Mr. Starbuck, they receive a ‘biography’ that is outrageously false," the lawsuit said.
"Google’s lies about Mr. Starbuck are not simply glitches: when probed, Gemini admitted that it was deliberately engineered to damage the reputation of individuals with whom Google executives disagree politically, including Mr. Starbuck."
Starbuck later discussed the allegations on "The Robby Starbuck Show”, saying that the company’s AI products have produced defamatory answers to people who ask questions about him since 2023.
According to the lawsuit, the fabricated claims included “elaborate rape allegations”, sexually assaulting a minor, flying on Jeffrey Epstein’s plane, as well as a lengthy criminal record including stalking, drug charges, and resisting arrest.
Starbuck also alleges that Gemini told him the fabricated claims had been shown to 2,843,917 unique users.
The lawsuit seeks more than $15 million in damages.
In response to the allegations, Google said it was not able to replicate the results in its consumer products. The company leaned on the outputs being “hallucinations” from the Bard AI.
"Most of these claims relate to hallucinations in Bard that we addressed in 2023. Hallucinations are a well-known issue for all LLMs, which we disclose and work hard to minimize. But, as everyone knows, if you’re creative enough, you can prompt a chatbot to say something misleading," Google spokesperson José Castañeda said.
The tech giant moved to dismiss the case, which would require Google to prove that the plaintiff would lack a legal claim even if their allegations were conceivably true.
Judge Meghan Adams ruled that Starbuck "pleads sufficient facts indicating the Outputs were published to third parties”.
"The Court views these allegations as sufficient to support a claim for emotional distress damages at this stage in the proceedings, being sufficiently well-pled to proceed to discovery and flesh out the allegations," the court filing says.
Stay updated with our latest stories and follow us on social media
Be the first to discover new stories, ideas, and updates from our team.
In a post on X, Starbuck declared “a major victory in court against Google and a major victory for humanity.”
The ruling, however, is not yet a final victory. Google will have to defend its case in court – and face a question that hasn’t yet been fully resolved.
In the US, a person can't win a defamation case just because a statement is false. They have to prove actual malice. How that standard applies when allegedly defamatory statements are generated by AI, which doesn’t have beliefs and doesn't "know" something is false, remains an open question.
In any case, the lawsuit could create serious implications for AI companies if it helps define when they can be held liable for defamatory content generated by their models.
