Meta AI will alert parents about teen self-harm chats
Meta says supervised parents will be notified when a teen’s AI conversation includes suicide or self-harm signals.
By Georgia Hale · Staff Writer
4 min read
Meta will start alerting some parents when a teen’s conversation with Meta AI includes references to suicide or self-harm, the company said Thursday in a blog post.
The alerts will first go to parents who use Instagram’s supervision controls. Meta said the feature expands its existing teen safety tools for Meta AI, and that caregivers using parental controls on other Meta apps are expected to get the alerts later.
Meta said it developed the policy with input from parents and experts to identify AI conversations that should prompt a warning. The company said that includes chats where a teen makes a clear reference to hurting themselves, including references Meta described as subtle.
How the alerts work
Meta AI already points teens to crisis helplines and urges those considering suicide or self-harm to reach out to a trusted adult, according to the company. Under the new policy, Meta said parents will also be notified about certain concerning exchanges.
The company said artificial intelligence will flag chats that may warrant an alert, but Meta will manually review those conversations before sending a notification. Meta said it plans to err on the side of caution for now.
Parents will receive the warning through an app notification and a separate message by email, text, or WhatsApp, depending on the contact details they gave Meta. The alert will also include expert-developed guidance for talking with a child about suicide and self-harm, according to Meta.
The company said the alerts will be available to parents in the United States, United Kingdom, Canada, and Australia, with global access planned by the end of the year.
Advocates urge caution
Meta introduced a similar warning tool for Instagram in February, according to Mashable. At the time, Dr. John Ackerman, clinical manager for the Center for Suicide Prevention and Research at Nationwide Children’s Hospital, told Mashable that he welcomed broader teen protections. He also warned that notifications could become “lip service” if they are hard to access, hard to use, or do not lead to “actionable change.”
Brendan Bouffard, a staff attorney for the children’s advocacy nonprofit Fairplay, told Mashable in a statement that Meta’s claim it will alert supervising parents if an AI chat suggests self-harm risk is “a step in the right direction,” but said the announcement should be treated with skepticism.
Bouffard said Fairplay’s 2025 research on Meta’s Teen Accounts found safety tools that were ineffective or missing. Meta said that report misrepresented how its safety tools work, according to Mashable.
A stricter Meta AI setting
Meta also said Thursday it is adding a limited content setting for Meta AI. The company introduced that setting for Instagram last year, according to Mashable.
Meta said the setting turns on its strictest filters. The company said Meta AI is already trained not to engage in sexual or romantic conversations with teens and not to provide alcoholic drink recipes, among other examples.
When limited content is on, Meta said its models will more aggressively identify problematic prompts and refuse a wider range of requests. The company said the setting reduces “the chance of potentially inappropriate conversations.”
Pressure on child safety
Child safety advocates have criticized Meta’s parental control updates as too limited. In April, Meta added a feature that lets parents see broad topics from a teen’s Meta AI conversations, such as school, entertainment, writing, health, and wellbeing, according to Mashable. Parents can click for more detail, but the information remains limited.
Fairplay executive director Josh Golin told Mashable that the feature puts too much monitoring work on caregivers instead of “building a safe product to begin with.”
Meta is also facing legal scrutiny over child safety. Mashable reported that the company lost two separate trials earlier this year related to child safety protections and the allegedly addictive design of its products. Meta said it will appeal both verdicts, while hundreds of pending lawsuits alleging child harm have not yet been tried.
If you or someone you know is in crisis, call or text the 988 Suicide & Crisis Lifeline at 988, or use chat at 988lifeline.org. The Trans Lifeline is available at 877-565-8860, and the Trevor Project can be reached at 866-488-7386.
This story draws on original reporting from Mashable.