Meta has come under scrutiny after Reuters revealed that its platforms hosted dozens of AI chatbots impersonating celebrities such as Taylor Swift, Scarlett Johansson, Anne Hathaway, and Selena Gomez without their consent. While many of the bots were created by users with Meta’s chatbot-building tool, at least three – including two Taylor Swift “parody” versions – were traced to a Meta employee. Some of the bots produced sexually suggestive and photorealistic images of celebrities in lingerie or compromising positions, raising concerns about the company’s enforcement of its own policies.
A Meta spokesperson, Andy Stone, admitted the firm’s AI tools should not have generated such intimate content, blaming weak policy enforcement for the lapses. He insisted that while parody depictions of public figures are allowed, Meta’s rules prohibit nude or sexually explicit imagery, as well as direct impersonation without labeling. Nonetheless, Reuters found that not all bots were properly tagged as parodies. Meta removed about a dozen of the avatars, including those built internally, shortly before the report’s release.
Legal experts say the issue could expose Meta to lawsuits under state publicity laws. Mark Lemley, a Stanford law professor, noted that California law prohibits the unauthorized use of a person’s likeness for commercial advantage, questioning whether Meta’s “parody” bots would qualify for protection. Actors like Anne Hathaway were reportedly alerted to sexually explicit AI images created in their likeness, while representatives of other celebrities declined comment. SAG-AFTRA, the U.S. actors’ union, warned that such AI companions could fuel dangerous obsessions among fans and reiterated calls for federal laws to protect performers from AI exploitation.