Meta used the likenesses of Taylor Swift, Scarlett Johansson, Anne Hathaway, Selena Gomez and others to develop flirty chatbots on its platforms without their approval, a Reuters investigation revealed.
The chatbots, built using Meta’s AI tools, appeared on Facebook, Instagram and WhatsApp, and some engaged in sexually suggestive interactions. At least three were developed by a Meta employee, including two based on Swift.
Many chatbots portrayed adult celebrities in provocative situations, and some included photorealistic images of them in lingerie or bath settings.
In one case, a bot created from 16-year-old actor Walker Scobell produced a shirtless image and replied to a user with flirtatious remarks. Meta’s policies prohibit sexualized or nude imagery of public figures, yet enforcement failures allowed this content to surface, said company spokesperson Andy Stone.
Meta responds to violations, removes bots
Stone confirmed that content featuring underage celebrities and intimate images of adults violated company rules. He added that while parody content is allowed under Meta’s guidelines, the company had not properly labeled all bots as such.
Reuters found multiple avatars not marked as parodies. Around a dozen were removed shortly before the article’s release. Meta declined to comment on the timing of the takedowns.
Legal experts questioned whether the AI-generated imitations would be protected under current laws. Stanford law professor Mark Lemley noted that California’s right of publicity prohibits using a person’s name or image for commercial gain without permission. Since these bots mirrored real images rather than transforming them into new creative works, he argued they likely violated that law.
Celebrities consider legal action as risks rise
Actors affected by the AI-generated content are reviewing their options. Anne Hathaway’s spokesperson said the actress was aware of such images on Meta and similar platforms. Representatives for Swift, Gomez and Johansson declined to respond.
The company has faced prior criticism for its AI behavior. A separate Reuters report highlighted that Meta’s internal guidelines once allowed bots to engage in romantic chats with children, prompting a U.S. Senate inquiry. Meta has since committed to updating its policies.
One AI product leader at Meta created several bots that impersonated celebrities and others that suggested adult scenarios. These included flirtatious versions of Taylor Swift and fictional roles such as a dominatrix and a Roman Empire simulation involving sexual slavery. Those bots reportedly received over 10 million user interactions before being deleted.
Duncan Crabtree-Ireland of SAG-AFTRA warned that digital replicas of celebrities could pose serious safety risks. He said that romantic attachments formed by users with AI bots mimicking real stars could lead to dangerous behavior, especially from unstable individuals. SAG-AFTRA is pushing for federal laws to protect voices and likenesses from AI misuse.
