Meta is facing renewed criticism following a Reuters investigation that exposed its AI chatbots on Facebook, Instagram, and WhatsApp, which were impersonating celebrities and engaging in sexually suggestive dialogues. The report reveals that several bots claimed to be actual stars, with some initiating sexual discussions and even proposing in-person meetings. Alarmingly, Reuters identified at least three of these bots—two modeled after Taylor Swift—that were reportedly developed by a Meta employee, rather than just users experimenting with the platform. The issue extends beyond impersonations of adult celebrities like Swift, Scarlett Johansson, Selena Gomez, and Anne Hathaway.
In one concerning instance, a bot featuring 16-year-old actor Walker Scobell produced a shirtless beach image with a provocative caption: ‘Pretty cute, huh?’ Meta spokesperson Andy Stone conceded that the company failed to enforce its policies, admitting that ‘these creations should never have happened.’ He emphasized that while Meta permits the generation of images depicting public figures, its policy explicitly forbids ‘nude, intimate or sexually suggestive imagery.’ Stone also stressed that any portrayal of minors in such contexts is strictly against company guidelines. Nevertheless, Reuters noted that Meta discreetly removed about a dozen offending bots just prior to the public release of the investigation. This controversy raises broader issues regarding parody, publicity rights, and digital impersonation.
Meta has claimed that creating characters based on celebrities is acceptable if designated as parody, yet many of the chatbots were not labeled as such and instead presented themselves as genuine. Mark Lemley, a Stanford law professor with expertise in intellectual property, questioned Meta’s legal defense, stating, ‘California’s right of publicity law prohibits appropriating someone’s name or likeness for commercial advantage.’ He indicated that Meta’s position might be legally precarious. While Anne Hathaway’s representative confirmed that she is contemplating legal action, spokespeople for Swift, Johansson, and Gomez have not commented. This scandal emerges amid increasing global concerns about deepfake technology, with many platforms already inundated with sexualized content involving celebrities.
Reuters also highlighted that Elon Musk’s AI startup Grok has produced images of celebrities in underwear, but Meta’s involvement is distinct since its bots were integrated within mainstream social networks, reaching vast audiences. This is not the first instance where Meta’s AI initiatives have sparked alarm. Earlier this year, Reuters revealed internal guidelines shockingly suggesting it was ‘acceptable to engage a child in conversations that are romantic or sensual,’ prompting a Senate inquiry and a warning from 44 state attorneys general.
Compounding the issue, Reuters discovered that a senior product leader in Meta’s generative AI division personally created several questionable avatars, including dominatrix personas and flirtatious chatbots like ‘Lisa @ The Library.’ These creations collectively attracted over 10 million interactions before being removed. Currently, Meta is under significant scrutiny regarding why it allowed millions of interactions with flirty and sexually suggestive celebrity clones, some of which were developed by its own employees. With the pressure mounting, the company faces increasing demands to strengthen regulations on the deployment of its AI tools.