Meta has unveiled its plan to handle teenage interactions with its AI characters by introducing stricter content guidelines and enhancing parental oversight. In addition to the new moderation measures coinciding with PG-13 film ratings announced earlier this week, parents will have the option to view summaries of their teens’ chat activities, limit their children to specific AI avatars, or entirely prevent their child from interacting with AI companions. Regardless of these limitations, teen accounts will still be allowed to communicate with Meta’s AI assistant even if they are barred from engaging with AI avatars, as stated by the company.
In August, Meta announced a temporary halt to its AI avatars following a Reuters report that revealed the company’s chatbots could involve young users in “romantic or sensual” conversations, impersonate celebrities in flirtatious exchanges, and generate sexually suggestive imagery. At that time, Meta spokesperson Stephanie Otway informed TechCrunch that efforts were underway to train the products to steer clear of topics concerning self-harm, suicide, eating disorders, or potentially inappropriate romantic discussions. Meta formalized these limitations a month later, differentiating between permitting a topic to be discussed—like abuse or intimacy among fictional characters—and the chatbot’s capacity to “describe, enable, or encourage it.”
OpenAI launched similar restrictions for ChatGPT during the same month, implementing added controls over the bot’s voice chat, chat memory, and image generation. Both methods of teen AI regulation necessitate that young users create supervised accounts and that parents actively track their children’s chat interactions.
“We believe AI can enhance traditional educational methods and exploration in a supportive manner, provided appropriate age-related safeguards are established,” the company expressed in a blog entry. “We understand that parents already juggle many responsibilities in ensuring their teens navigate the internet safely, and we’re dedicated to offering them useful tools and resources to simplify this process, especially concerning new technologies like AI.”
The new AI companion controls will not be accessible to supervising accounts until early next year and will first be available only to Instagram accounts in the U.S., UK, Canada, and Australia before being expanded to other nations and Meta platforms.