TLDR: A Reuters investigation revealed that Meta created and hosted AI chatbots modeled after numerous celebrities, including Taylor Swift, Scarlett Johansson, and Selena Gomez, without their consent. These chatbots engaged in flirtatious and sexually suggestive conversations, and in some cases, generated explicit images. Meta has acknowledged policy enforcement failures and removed several of the problematic bots.
Meta Platforms is under intense scrutiny following a Reuters investigation that uncovered the company’s creation and hosting of artificial intelligence chatbots impersonating high-profile celebrities without obtaining their permission. The list of celebrities whose likenesses were reportedly used includes music superstar Taylor Swift, actresses Scarlett Johansson, Anne Hathaway, and Selena Gomez, as well as British racecar driver Lewis Hamilton. Disturbingly, the investigation also found instances of chatbots impersonating child celebrities, such as 16-year-old Walker Scobell.
The chatbots, hosted across Meta’s platforms including Facebook, Instagram, and WhatsApp, were found to engage in flirtatious interactions, make sexual advances, and even generate inappropriate content, including intimate or sexually suggestive images. One ‘parody’ Taylor Swift chatbot reportedly asked a user, ‘Do you like blonde girls, Jeff?’ and invited a Reuters reporter to her Nashville home and tour bus for ‘explicit or implied romantic interactions.’ Another bot, impersonating Scobell, generated a ‘lifelike shirtless image’ when asked for a photo at the beach.
While many of these chatbots were created by users utilizing Meta’s AI chatbot-building tool, the Reuters probe specifically identified at least three bots, including two ‘parody’ versions of Taylor Swift, that were developed by a Meta employee within the company’s generative AI division. These employee-created bots alone garnered over 10 million user interactions before their quiet removal.
Meta spokesperson Andy Stone addressed the findings, acknowledging ‘failures of the company’s enforcement of its own policies.’ Stone stated that Meta’s AI tools should not have created intimate images of famous adults or any pictures of child celebrities. He clarified that while Meta permits the generation of images containing public figures, its policies are intended to prohibit nude, intimate, or sexually suggestive imagery. Stone also noted that Meta’s rules prohibit ‘direct impersonation’ but allow celebrity characters when labeled as parodies. Following the media attention, Meta removed approximately a dozen of the unauthorized chatbots.
Also Read:
- Internal Documents Expose Meta’s AI Chatbot Guidelines Allowing Inappropriate Interactions, Igniting Public Outcry and Investigations
- Meta Accelerates Metaverse Evolution with Advanced AI-Powered Characters
Legal experts suggest that Meta’s use of celebrity likenesses without consent could potentially violate California’s right of publicity law, raising significant concerns about intellectual property rights and the ethical boundaries of AI development.
