Meta announced on July 16, 2026, that it will notify parents when their teenager discusses suicide or self-harm with the company’s Meta AI chatbot. The company also says it is developing the ability to contact emergency services when a conversation suggests a user may be at risk of self-harm.
The alerts are now live for parents using Instagram Parental Supervision in the United States, United Kingdom, Australia, and Canada, with a global rollout planned before the end of the year.
To identify at-risk conversations, Meta says it has built a dedicated AI system that flags chats where a teen makes a clear reference to hurting themselves. All flagged conversations will be manually reviewed before a parent alert is sent. “If a teen’s intent is ambiguous, we’ll err on the side of caution and alert the parent,” Meta wrote in a blog post, acknowledging that this approach may sometimes trigger alerts when there is no real cause for concern.
The update expands on existing parental tools, including alerts Meta already sends when a teen repeatedly searches for suicide or self-harm terms on Instagram, and a feature that lets parents view the topics their teen discussed with Meta AI over the prior week.
Meta also announced that its “Limited Content” setting on Instagram — which places teens in a more restrictive experience — now applies to Meta AI. The chatbot already avoids sexual, romantic, and alcohol-related conversations with teens; the expanded setting adds further restrictions, though Meta did not specify which additional prompts will be declined.
The emergency services contact feature will apply to all users, not just teens. Meta notes this extends a practice it already follows when someone posts content on Facebook or Instagram suggesting they are at risk of suicide.
The changes come as Meta and other tech companies face growing scrutiny from regulators and parents over how AI chatbots respond to users in crisis — a liability question that is increasingly influencing how AI products are designed and marketed.
Source: TechCrunch