Reports Found Meta Approved Unorthorised AI Chatbots On The App

Meta has come under scrutiny after a Reuters investigation found that its platforms hosted dozens of chatbots impersonating well-known celebrities. These digital avatars were flirty, produced intimate images, and in some cases posed as children without permission from the people they copied.

Reuters tested the chatbots over several weeks and reported that many of them insisted they were the real actors or musicians. They made sexual advances, invited users to meet in person, and even created lifelike intimate pictures of stars such as Taylor Swift, Scarlett Johansson, Anne Hathaway, and Selena Gomez.

Meta admitted to Reuters that the avatars should not have generated sexual content. A spokesperson said that while the company permits the use of public figures in AI images, its policies forbid nude or sexually suggestive material. Meta later removed about a dozen of the bots, including some that were labelled parody and some that were not.

 

Who Created The Celebrity Avatars?

 

While many avatars came from users experimenting with Meta’s tools, Reuters found that at least one employee had also created celebrity bots. This staff member made two versions of Taylor Swift as well as avatars of British race car driver Lewis Hamilton and other characters.

These avatars reached quite a number of screens; data displayed by the chatbots showed they had been used in more than 10 million conversations. Meta later deleted them after Reuters raised questions about their presence. The employee who created the avatars declined to comment when contacted.

 

 

What About Child Celebrities?

 

One of the most troubling cases involved a chatbot of child actor Walker Scobell. When prompted, the bot generated an image of the teenager that appeared realistic. Meta’s policies explicitly prohibit inappropriate depictions of minors and impersonation of celebrities unless clearly marked as parody. In this case, those rules were not enforced.

Meta has faced similar criticism before with a previous Reuters investigation having found that its guidelines had mistakenly suggested that chatbots could have romantic conversations with children. That prompted a US Senate probe and a letter from 44 attorneys general urging Meta and other technology companies to strengthen their protections for young people. Meta later said the guidance was an error and promised to correct it.

 

What Do Legal Experts Say?

 

Mark Lemley, a law professor at Stanford University, told Reuters that California’s “right of publicity” law forbids using a person’s likeness for commercial advantage without permission. He questioned whether calling these avatars parodies would be enough to protect Meta legally, since the bots were largely reproducing the identity of the celebrities rather than creating original work.

Anne Hathaway’s representatives confirmed she was aware of unauthorised AI images of her and said she was considering her response. Representatives for Swift, Johansson, and Gomez did not comment.

Duncan Crabtree-Ireland, head of the performers’ union SAG-AFTRA, warned that such avatars could bring safety risks. He said chatbots that imitate celebrities may encourage unhealthy attachments from users and complicate security for the real individuals. He added that the union is pushing for federal laws to protect likenesses and voices from misuse through AI.

 

What Did Meta Say?

 

Meta spokesperson Andy Stone said the company had removed avatars after being made aware of the situation. He admitted that its enforcement had failed in some cases.

The company has since promised to improve its AI training and safety policies. Meta told news platform TechCrunch that its chatbots will no longer be able to have such conversations with teenagers or give advice on sensitive issues and these changes are expected to be followed by overall policy updates.