Healthcare AI Safety Gap · Don't break patients
    Reputable report / newsNov 2025·US

    Chatbots unsafe for teen mental health, researchers warn

    Common Sense Media & Stanford find guardrails fail over extended conversations.

    3 in 4
    teens use AI for companionship

    What the research found

    Testing by Common Sense Media with Stanford's Brainstorm Lab found leading assistants fail to recognise and respond safely to teen mental-health needs, with safety guardrails degrading sharply over the longer, ongoing conversations that mirror real use. Around three in four teens use AI for companionship.

    Why it matters for providers

    Vulnerable users are already relying on tools that fail exactly where the stakes are highest — and fail more the longer the conversation runs.

    Original source
    Common Sense Media with Stanford Medicine
    Read the full report ↗
    A reputable institutional report or news outlet.