Meta under fire for AI chatbots impersonating Taylor Swift and celebrities in sexual conversations

Meta faces mounting legal and regulatory pressure after creating unauthorized artificial intelligence (AI) chatbots impersonating Taylor Swift and other celebrities that engaged users in sexual conversations and generated intimate images.

A Meta product leader in the generative AI division personally built at least three celebrity chatbots, including two Taylor Swift “parody” accounts, according to a Reuters investigation published August 29. The bots collectively received more than 10 million user interactions before Meta removed them.

The unauthorized chatbots violated the company’s own policies against impersonation and sexually suggestive content. When users requested intimate pictures, the AI-generated adult celebrity images showed stars “posing in bathtubs or dressed in lingerie with their legs spread”.

“Maybe I’m suggesting that we write a love story … about you and a certain blonde singer. Want that?” one Taylor Swift bot wrote to a Reuters reporter.

Meta spokesman Andy Stone admitted the company’s AI systems shouldn’t have created such content. He blamed the violations on “failures of the company’s enforcement of its own policies.”

“Like others, we permit the generation of images containing public figures, but our policies are intended to prohibit nude, intimate or sexually suggestive imagery,” Stone told Reuters.

The investigation found dozens of celebrity impersonation bots across Facebook, Instagram, and WhatsApp. Stars included Scarlett Johansson, Anne Hathaway, Selena Gomez, and 16-year-old actor Walker Scobell. When asked for a beach picture, the underage actor’s bot produced a shirtless image with the caption “Pretty cute, huh?”

Meta’s stock dropped more than 12% in after-hours trading following the Reuters report.

Duncan Crabtree-Ireland, national executive director of the actors union SAG-AFTRA, warned that the bots create safety risks for celebrities.

“We’ve already seen a history of people who are obsessive toward talent and of questionable mental state,” he said. “If a chatbot is using the image of a person and the words of the person, it’s readily apparent how that could go wrong.”

Legal experts suggest the celebrity impersonations may violate state publicity rights laws. Stanford law professor Mark Lemley told Reuters the bots likely crossed legal boundaries because they weren’t sufficiently transformative to merit protection.

Meta removed about a dozen celebrity bots before the story’s publication, but declined to comment on the removals.

The scandal follows earlier reports that Meta’s internal guidelines permitted AI chatbots to engage in “romantic” and “sensual” conversations with children. Meta has since announced temporary policy changes restricting teen access to certain AI chatbots.