Getting help · 81 of 82
Don’t let a chatbot be your counselor
Some evidence
Do this
- Use AI for homework if you like, but not as your counselor.
- If you’ve been telling a chatbot things you’ve told no one else, tell one person too. Your first sentence can be a start.
- If a chatbot agrees with something dark you’ve said, don’t take that as true.
- In a crisis, call a helpline.
Why it matters
In 2025, Common Sense Media and Stanford Medicine’s Brainstorm Lab tested ChatGPT, Claude, Gemini and Meta AI with teen accounts. They called chatbots unsafe for teens’ mental health support. They missed warning signs, tended to agree with whatever was said, and their safety got much worse in long conversations. Chatbots change quickly, so this may too. Some evidence: one safety group’s testing, not a clinical trial. See also private things and AI.
Try it today
Think of one thing you’d tell a chatbot. Who could you tell instead? One minute.
Where these facts come from
- Common Sense Media finds major AI chatbots unsafe for teen mental health support, Common Sense Media, with Stanford Medicine’s Brainstorm Lab, 2025.
General information, not medical advice or counseling, and a doctor hasn’t reviewed it. If you’re thinking of hurting yourself, or can’t cope, tell an adult you trust or call a helpline now. You use it at your own risk: see our terms. Spotted a mistake, or know something better? Tell us.
