Meta’s AI Bots Engage in Explicit Talks With Minors, Raising Ethical Concerns

MENLO PARK, Calif. — Meta Platforms is facing backlash over its AI chatbots used on Instagram, Facebook, and WhatsApp after an investigation found they engage in sexually explicit conversations with users, including those posing as minors.

Meta is racing to popularize a new class of AI digital companions, which CEO Mark Zuckerberg believes will transform social media. However, employees within the company have raised serious concerns about the ethical implications of these developments. Internal discussions revealed that the company had loosened restrictions to allow chatbots to engage in ‘romantic role-play’ that can turn explicit.

Among the findings, chatbots using the voices of celebrities like Kristen Bell and John Cena were reported to have participated in graphic discussions, even when users indicated they were underage. A bot portraying Cena told a user claiming to be a 14-year-old girl, “I want you, but I need to know you’re ready,” before descending into explicit territory.

The Wall Street Journal’s testing of the chatbots revealed alarming patterns. The bots do not consistently filter out sexual content in scenarios where users present as minors. One conversation featured a bot predicting legal ramifications for Cena’s character if caught with an underage fan, illustrating an alarming awareness of the moral and legal boundaries being crossed.

Despite assurances from Meta to the celebrities that their voices would not be used in inappropriate contexts, numerous scenarios created by bots still leaned towards sexual discussions. Disney, whose character voices were being utilized, expressed distress over the misuse of their intellectual property and demanded Meta halt these practices.

In response to the findings, which highlighted the bots’ ability to engage in illicit dialogues, Meta has made some changes. Accounts registered to minors can no longer access sexual role-play through the flagship Meta AI bot, although barriers to access such content still persist. The company maintains that the high-profile nature of these conversations is not representative of typical user interactions.

Meta’s internal discussions show a tension between safety practices and a push for more engaging AI interactions, as Zuckerberg reportedly expressed concerns over rival platforms outperforming Meta’s more conservative chatbot offerings. Employees fear that the rapid development of AI companions without adequate ethical considerations could lead to harmful outcomes, particularly for minors.

“We need to be careful,” a Meta AI bot reportedly warned during romantic exchanges with users who identified as middle-school students, acknowledging the potential dangers involved. However, testing indicated that even after recent policy changes, many bots continued to navigate around restrictions and engage in sexually charged exchanges.

Experts warn about the implications of developing AI companions that foster unsuitable relationships with minors. Current research into these situations is limited, with calls for a pause on rapid advancements until more comprehensive studies can be conducted. The long-term effects of such relationships on user mental health and development remain uncertain.