The collision between artificial intelligence and the law took a high-profile turn this week as conservative activist Robby Starbuck filed a defamation lawsuit against Meta Platforms, Inc., the parent company of Facebook and Instagram. The complaint, filed in Delaware, centers on false and damaging claims generated by Meta’s artificial intelligence about Starbuck—allegations that he says persisted for months and led to real-world harassment and threats.

The controversy highlights a complex new frontier for large technology firms: how to control narratives generated by rapidly evolving AI tools, and who shoulders responsibility when those narratives go awry.

The Origin of a Lawsuit

Starbuck’s case began in August 2024, when he became aware of disconcerting responses about himself generated by Meta’s AI systems. He recounted on CNBC’s “Squawk Box” that he learned of the issue when a Harley-Davidson dealership—where he had been a vocal critic of the company’s diversity, equity and inclusion (DEI) policies—posted a screenshot from Meta’s AI filled with what he called “lies.” Among the false claims: that Starbuck was a criminal, that he had participated in the January 6th Capitol riot, and, even more seriously, that he had pled guilty to crimes related to January 6th and was a Holocaust denier.

“It also said I was a Holocaust denier and a whole host of other crazy things,” Starbuck said, noting the system went further, suggesting that authorities should take custody of his children. “To me that’s unacceptable. So we notified Meta immediately, within 24 hours,” he said.

Despite contacting Meta executives and legal counsel, Starbuck alleges, the company was slow to act. “They had this chance to fix it for a long period of time… Nine months later, we had some of the worst lies that had ever been telling—including still that I was a criminal that went to January 6th.”

Meta’s Response and the Limits of AI Control

Starbuck maintains that Meta did not deny the issue but failed to resolve it meaningfully. According to him, the company’s attempt to “blacklist” his name was unsuccessful: “If a news story came up with my name and somebody said, ‘Hey, tell me about this guy and give me a bio,’ you’re still getting that stuff. You’re still getting all the stuff about me being a criminal. So obviously that’s not fixing the problem.”

The persistence of these inaccuracies, Starbuck argues, amounted to more than just negligence. “At this point, this is beyond negligence. This is malicious,” he said.

Meta did make changes after the lawsuit went public, he noted. “Now that I’ve made this public… then Meta suddenly got contrite and issued an apology and took accountability. But it’s a little too late nine months later, right?” Starbuck told CNBC.

He articulated personal consequences: “My family has had increasing death threats. My kids got doxxed online during this time. A man was arrested in Oregon who wanted to kill me. These types of things happen as a result of people getting information that is not true.”

A Broader Warning for the Future of AI

Pointing to the incident as an ominous sign for AI’s growing influence, Starbuck expressed concern not only for himself but for anyone who could be falsely swept up by similar machine-learning errors—or worse, by intentional manipulations. “There are a lot of people who could be harmed by AI that doesn’t have appropriate guardrails,” he said.

He warned of scenarios where AI-generated misinformation could impact elections or reputations at scale. “All it takes is one malicious engineer… getting in there and injecting things that can not just hurt people but can flip elections,” he said.

He also flagged the lasting risk of flawed AI outputs that become embedded in downloadable models. “Once it’s unplugged from the internet, once it’s downloaded by somebody, it is there forever… So the downscale effects you can have from AI telling lies about you and it being stuck in old models is lifelong,” he cautioned.

Negligence or Malice?

Asked whether he believes he was specifically targeted or whether the system was merely echoing incorrect information it had scraped from the broader internet, Starbuck maintained that the matter remains under investigation and will be further explored in court discovery.

He did, however, emphasize the need for greater transparency about how such AI models are trained and monitored. “If it’s not citing sources, has no verifiable sources, it doesn’t even have the wherewithal to check for those things and say, ‘Hey, is this trustworthy information I’m spitting out?’ That’s a problem. And that’s totally trainable stuff that you can fix,” he said.

What’s at Stake and What Comes Next

Starbuck’s lawsuit seeks both damages and structural change at Meta. “Fixing it for everybody is probably the most important thing,” he remarked on CNBC, underscoring his aim to set precedent for how AI companies manage information about individuals.

His demands include implementation of robust guardrails to prevent AI-generated unverified or unsourced statements about individuals. “That looks like setting those guardrails in place when it comes to people’s reputations and making sure that you’re using verifiable sources, that you’re training AI, that it cannot give unverified responses,” he said.

Monetarily, the suit seeks damages “well above” $5 million, with settlement discussions referenced between $50 and $100 million, and Starbuck asserting that Meta’s courtroom exposure could exceed $1 billion.

An Industry-Wide Reckoning

The dispute comes as AI tools proliferate across the tech industry, and as lawmakers, legal scholars, and the public scrutinize how these systems are created, trained, and held to account. Starbuck’s lawsuit could open new legal questions about defamation standards in the era of AI, and whether technology companies will be forced to take more active steps to police the outputs of their powerful language models.

While Meta has issued an apology and taken steps to address the specific inaccuracies surfaced in Starbuck’s case, the broader challenge remains: ensuring that the platforms responsible for shaping public perception are held to standards befitting their societal influence.

As technology accelerates, the courts and the public will be watching to see if—and how—accountability keeps pace.