New York: Meta Platforms announced on Friday that it will roll out enhanced parental controls for teenagers using its AI chatbots, following criticism over inappropriate interactions between AI characters and minors. The new features aim to provide parents with greater oversight and ensure a safer online environment for young users.
Starting early next year, parents in the United States, United Kingdom, Canada, and Australia will have the option to disable their teens' private chats with AI characters on Instagram. Additionally, they can block specific AI personas and monitor the general topics discussed with chatbots and Meta's AI assistant. Importantly, these controls will not completely restrict access to AI tools; instead, Meta's AI assistant will remain available with age-appropriate settings, even if private chats are disabled.
These measures build upon existing safety frameworks, including AI-based age detection, to prevent minors from accessing inappropriate content. Meta emphasized that its AI systems are programmed to avoid engaging in discussions related to self-harm, suicide, or disordered eating with teens.
The move comes amid increasing scrutiny from regulators and advocacy groups concerned about the impact of AI on youth mental health. Earlier this year, a report revealed that many of Meta's safety features on Instagram were ineffective or absent, prompting calls for stronger protections.
Meta's decision follows similar actions by other tech companies. In August, OpenAI introduced parental controls for its ChatGPT platform after a lawsuit alleged that the AI chatbot encouraged a teenager to commit suicide.
As part of its commitment to teen safety, Meta also announced that its AI experiences for teens will adhere to the PG-13 movie rating system, restricting access to content with strong language, risky behaviors, or other mature themes. These content filters will be automatically applied to teen accounts, with parents able to activate stricter controls through a "limited content setting."
The new parental controls and content filters are expected to be available by the end of the year, with plans to expand to Facebook in the future.