Meta will notify parents about dangerous conversations between teenagers and AI
Meta is introducing a new safety feature that will notify parents if their teens discuss topics related to suicide or other serious risks to life and health with AI-powered chatbots.
The new “Notifications for Parents” feature will begin rolling out on Thursday to users of Instagram’s parental controls in the U.S., the U.K., Australia, and Canada. The company plans to make it available worldwide by the end of the year.
Meta reported that, along with the notification, parents will receive informational materials developed in collaboration with experts to help them talk to teens about suicide and self-harm.
According to the company, a specialized artificial intelligence system for detecting potentially dangerous conversations was created after consulting with parents and mental health experts to determine which types of interactions should trigger a warning. As an additional safety measure, every conversation flagged by the AI system will be reviewed by a human before a notification is sent. Meta also noted that if a teenager’s intentions remain unclear, the company will “err on the side of caution.”
According to current guidelines, the minimum age for using Meta’s social media platforms—Facebook and Instagram—is 13. The company notes that for users under 18, content settings for “teen accounts” (13+) are automatically applied, which also extend to interactions with Meta AI.
In addition, Meta announced that it is developing a feature to contact emergency services if a conversation with Meta AI indicates an immediate risk of suicide.
To improve Meta AI’s response to queries related to suicide and self-harm, the company sought feedback from experts specializing in adolescent mental health.
Last year, OpenAI announced the introduction of parental controls with a notification feature for teenagers. In addition, the company reported that it uses an age-estimation model in ChatGPT that automatically applies a special teen mode to users under 18. This mode excludes ads and includes additional safeguards against content that is sensitive or inappropriate for their age group. Currently, according to OpenAI, ChatGPT is available to children aged 13 and older, and children between the ages of 13 and 18 must obtain parental consent before using ChatGPT.
This is reported in an ABC News article.
We should add that OpenAI is shutting down its AI browser, Atlas, less than a year after its launch.
As a reminder, Microsoft plans to allocate approximately $80 billion in fiscal year 2025 to develop data centers for training artificial intelligence models.
Tesla says it will build new “first-of-their-kind” data centers. The automaker is hiring staff for this purpose and acquiring some existing data centers.
