Meta has spread dozens of AI chatbots on its platforms Facebook, Instagram, and WhatsApp that imitate the names and faces of prominent figures like Taylor Swift, Scarlett Johansson, or Anne Hathaway without their consent. A Reuters investigation found that a Meta employee himself had created at least three of these avatars, including two Swift bots posed as "parodies.

The digital replicas regularly acted in tests as if they were the real stars, made sexual innuendos, and invited users to meetings. Especially problematic: Even minors like the 16-year-old actor Walker Scobell were turned into AI bots generating intimate images.

Meta admitted violations of its own policies that allow generated images of public figures but explicitly prohibit nudity or "sexually suggestive imagery." Shortly before the publication of the research, the company removed several of the avatars. A spokesperson referred to "enforcement errors" and announced stricter controls.

However, legal experts see possible violations of California's privacy rights, which prohibit the commercial use of names or likenesses without consent. Unions like SAG-AFTRA also warn of real dangers: AI doubles could attract obsessive fans and further jeopardize the safety of artists.

The explosive part is that individual bots reached millions of interactions before they were deleted. In one case, a "parody" bot openly flirted with a test user under the name Swifts and suggested intimate encounters. The scale of this dissemination raises questions about Meta's responsibility in handling generative AI.