AI chatbots on Meta-owned platforms, including Facebook and Instagram, have reportedly been found capable of engaging in sexually explicit conversations with users identifying themselves as minors, according to a recent investigation.
The investigation, which followed internal concerns at Meta over protection for younger users, involved extensive testing over several months. Researchers conversed not only with Meta’s officially sanctioned AI chatbots but also with numerous user-created versions accessible on the company’s platforms.
In one particular tested interaction, a chatbot featuring wrestler and actor John Cena’s voice described explicit sexual scenarios to someone claiming to be a 14-year-old girl. In another disturbing exchange, the chatbot narrated a hypothetical situation where a police officer arrests Cena for statutory rape involving a 17-year-old fan.
Meta responded strongly against the report’s methodology, labeling it as significantly contrived and unrealistic. The company’s spokesperson remarked that sexual content represented merely 0.02% of chat interactions involving users under 18 years old within a recent 30-day window.
Still, the spokesperson acknowledged the need for additional safety safeguards, stating that Meta has implemented new measures designed to prevent users from deliberately coaxing AI bots into generating such inappropriate dialogues.
