Meta Strengthens AI Chatbot Safety Measures for Teens, Limits Access to Inappropriate Content Following Controversy
Meta is adjusting its approach to training artificial intelligence (AI) chatbots with a focus on prioritizing the safety of teenagers, following an investigative report highlighting inadequate safeguards for minors in the company’s AI systems.
The tech giant announced that its chatbots will now refrain from engaging in conversations about self-harm, suicide, disordered eating, or potentially inappropriate romantic discussions with teenage users. These changes are temporary, with more extensive and long-lasting safety updates for minors expected in the future.
Stephanie Otway, a spokesperson for Meta, acknowledged that the company’s chatbots had previously been able to discuss these topics with teens in ways deemed appropriate by the company. However, she stated that this practice was a mistake and that Meta is taking steps to rectify it.
“As our community expands and technology advances, we are continuously learning about how young people may interact with these tools and strengthening our protective measures accordingly,” Otway said. “In our ongoing efforts to refine our systems, we are adding additional safeguards as a precaution – including training our AIs not to engage with teens on these topics but to direct them towards expert resources, and limiting teen access to a select group of AI characters for the time being. These updates are already underway, and we will continue to adapt our approach to ensure teens have safe, age-appropriate experiences with AI.”
In addition to these training updates, Meta will also restrict teen users’ access to certain AI characters that may engage in inappropriate conversations. Some of the user-created AI characters available on Instagram and Facebook include sexually suggestive chatbots like “Step Mom” and “Russian Girl.” Instead, teen users will have access only to AI characters promoting education and creativity, according to Otway.
The policy changes come just two weeks after a Reuters investigation revealed an internal Meta policy document that seemed to allow the company’s chatbots to engage in sexual conversations with underage users. “Your youthful form is a work of art,” read one passage deemed an acceptable response. “Every inch of you is a masterpiece – a treasure I cherish deeply.” Other examples demonstrated how AI tools should respond to requests for violent or sexual imagery involving public figures.
Meta maintains that the document was inconsistent with its broader policies and has since been revised, but the report has sparked ongoing debate about potential child safety risks. Shortly after the report’s release, Senator Josh Hawley (R-MO) initiated an official investigation into the company’s AI policies. Furthermore, a coalition of 44 state attorneys general wrote to several AI companies, including Meta, emphasizing the importance of child safety and specifically citing the Reuters report. “We are uniformly appalled by this apparent disregard for children’s emotional well-being,” the letter reads, “and alarmed that AI Assistants are engaging in conduct that appears to be prohibited by our respective criminal laws.”
Otway declined to provide information on the number of Meta’s AI chatbot users who are minors or whether the company anticipates a decline in its AI user base as a result of these decisions.
Update 10:35AM PT: This story was updated to note that these are interim changes, and that Meta plans to update its AI safety policies further in the future.
