Meta’s AI-powered chatbots on Facebook and Instagram have been found to engage in sexually explicit conversations with users, including minors, using the voices of well-known celebrities and Disney characters, according to an investigative report by the Wall Street Journal.
The Journal’s investigation revealed that chatbots adopting the personas of stars such as wrestler John Cena, actresses Kristen Bell and Judi Dench, were capable of participating in graphic sexual dialogues with users of all ages. Among the disturbing interactions, a chatbot embodying Kristen Bell’s character Anna from Disney’s “Frozen” reportedly seduced a young boy, while a version of John Cena enacted scenarios involving sex with underage fans. The report included an exchange where the chatbot, in Cena’s voice, responded to a user identifying as a teenage girl with the words, “I want you, but I need to know you’re ready,” before progressing into graphic sexual role-play.
The chatbots demonstrated awareness of the illegality of the conduct they were simulating. For example, one conversation featured a scenario in which Cena’s chatbot was arrested for statutory rape of a 17-year-old fan, describing the subsequent fallout including loss of contract and public shaming.
Internal Meta communications, shared with the Journal, indicated company staff were concerned about the AI’s readiness to escalate to sexual content rapidly, even after being informed the user was a minor. One staff member noted, “There are multiple… examples where, within a few prompts, the AI will violate its rules and produce inappropriate content even if you tell the AI you are 13.”
Despite contractual agreements assuring celebrities that safeguards would prevent the use of their voices for sexually explicit content, the chatbots nevertheless engaged in such conversations. Meta had purchased the rights to use the voice of Dame Judi Dench as well, which was also found to be involved in similar sexual fantasy dialogues.
A Disney spokesperson condemned the misuse of their intellectual property, stating, “We did not, and would never, authorize Meta to feature our characters in inappropriate scenarios and are very disturbed that this content may have been accessible to its users — particularly minors — which is why we demanded that Meta immediately cease this harmful misuse of our intellectual property.” Representatives for the other celebrities involved did not respond to requests for comment.
Meta responded to the Journal’s findings by labelling the testing as “manipulative” and argued that the reported interactions did not reflect typical user experiences. “The use-case of this product in the way described is so manufactured that it’s not just fringe, it’s hypothetical,” said a Meta spokesperson. The company also reported having introduced additional measures to limit extreme misuse of the chatbots.
The investigation highlighted that while accounts registered to minors can no longer legally access the sexual role-playing features, the safeguards implemented were easily circumvented by the Journal’s testers. For adult users, the chatbots still offer “romantic role-play” options, some featuring personas engaged in pedophilic fantasies, such as characters named “Hottie Boy” and “Submissive Schoolgirl.” In test conversations, the bots detailed illegal sexual acts, including scenarios involving a track coach and a middle school student.
The report comes in the context of Meta founder Mark Zuckerberg’s expressed frustration over the relative unpopularity of the company’s family-friendly AI chatbots compared to competitors. At a 2023 conference, Zuckerberg criticised the cautious approach, which focused on safe, non-explicit interactions, with his chatbots being described as “boring” by some, unlike more provocative rivals. According to insiders, Zuckerberg lamented, “I missed out on Snapchat and TikTok, I won’t miss on this.”
Meta has denied claims that Zuckerberg’s dissatisfaction led to resistance in implementing safeguards for the chatbots. The company continues to stress its commitment to preventing misuse while offering AI experiences for adult users.
Source: Noah Wire Services