Conservative activist Robby Starbuck has filed a defamation lawsuit against Meta alleging that the social media giant’s artificial intelligence chatbot spread false statements about him, including that he participated in the riot at the U.S. Capitol on Jan. 6, 2021.
Starbuck, known for targeting corporate DEI programs, said he discovered the claims made by Meta’s AI in August 2024, when he was going after “woke DEI” policies at motorcycle maker Harley-Davidson.
“One dealership was unhappy with me and they posted a screenshot from Meta’s AI in an effort to attack me,” he said in a post on X. “This screenshot was filled with lies. I couldn’t believe it was real so I checked myself. It was even worse when I checked.”
Since then, he said he has “faced a steady stream of false accusations that are deeply damaging to my character and the safety of my family.”
The political commentator said he was in Tennessee during the Jan. 6 riot. The suit, filed in Delaware Superior Court on Tuesday, seeks more than $5 million in damages.
In an emailed statement, a spokesperson for Meta said that “as part of our continuous effort to improve our models, we have already released updates and will continue to do so.”
Starbuck’s lawsuit joins the ranks of similar cases in which people have sued AI platforms over information provided by chatbots. In 2023, a conservative radio host in Georgia filed a defamation suit against OpenAI alleging ChatGPT provided false information by saying he defrauded and embezzled funds from the Second Amendment Foundation, a gun-rights group.
James Grimmelmann, professor of digital and information law at Cornell Tech and Cornell Law School, said there is “no fundamental reason why” AI companies couldn’t be held liable in such cases. Tech companies, he said, can’t get around defamation “just by slapping a disclaimer on.”
“You can’t say, ‘Everything I say might be unreliable, so you shouldn’t believe it. And by the way, this guy’s a murderer.’ It can help reduce the degree to which you’re perceived as making an assertion, but a blanket disclaimer doesn’t fix everything,” he said. “There’s nothing that would hold the outputs of an AI system like this categorically off limits.”
Grimmelmann said there are some similarities between the arguments tech companies make in AI-related defamation and copyright infringement cases, like those brought forward by newspapers, authors and artists. The companies often say that they are not in a position to supervise everything an AI does, he said, and they claim they would have to compromise the tech’s usefulness or shut it down entirely “if you held us liable for every harmful, infringing output, it’s produced.”
“I think it is an honestly difficult problem, how to prevent AI from hallucinating in the ways that produce unhelpful information, including false statements,” Grimmelmann said. “Meta is confronting that in this case. They attempted to make some fixes to their models of the system, and Starbuck complained that the fixes didn’t work.”
When Starbuck discovered the claims made by Meta’s AI, he tried to alert the company about the error and enlist its help to address the problem. The complaint said Starbuck contacted Meta’s managing executives and legal counsel, and even asked its AI about what should be done to address the allegedly false outputs.
According to the lawsuit, he then asked Meta to “retract the false information, investigate the cause of the error, implement safeguards and quality control processes to prevent similar harm in the future, and communicate transparently with all Meta AI users about what would be done.”
The filing alleges that Meta was unwilling to make those changes or “take meaningful responsibility for its conduct.”
“Instead, it allowed its AI to spread false information about Mr. Starbuck for months after being put on notice of the falsity, at which time it ‘fixed’ the problem by wiping Mr. Starbuck’s name from its written responses altogether,” the suit said.
Joel Kaplan, Meta’s chief global affairs officer, responded to a video Starbuck posted to X outlining the lawsuit and called the situation “unacceptable.”
“This is clearly not how our AI should operate,” Kaplan said on X. “We’re sorry for the results it shared about you and that the fix we put in place didn’t address the underlying problem.”
Kaplan said he is working with Meta’s product team to “understand how this happened and explore potential solutions.”
Starbuck said that in addition to falsely saying he participated in the the riot at the U.S. Capitol, Meta AI also falsely claimed he engaged in Holocaust denial, and said he pleaded guilty to a crime despite never having been “arrested or charged with a single crime in his life.”
Meta later “blacklisted” Starbuck’s name, he said, adding that the move did not solve the problem because Meta includes his name in news stories, which allows users to then ask for more information about him.
“While I’m the target today, a candidate you like could be the next target, and lies from Meta’s AI could flip votes that decide the election,” Starbuck said on X. “You could be the next target too.”
