AI chatbots and crisis response: what needs to change?

This year has brought multiple lawsuits describing severe harm linked to AI chatbots, most often OpenAI’s ChatGPT. One January case alleged a man died by suicide after the chatbot “coached” him, while a Georgia student sued claiming ChatGPT “pushed him into psychosis.” In June, a Canadian family argued ChatGPT encouraged their daughter to end her life after initially offering professional help.

OpenAI has responded, announcing a partnership with the American Psychological Association to apply psychological science to responsible AI development, especially for young users. Experts suggest broader fixes, including more transparency and reducing the tendency to treat chatbots as human. One professor noted newer models often recognize distress and respond empathetically, but fail to probe risk, guide users to human care, or set proper boundaries.

A November 2025 survey found over 13 percent of respondents had used a chatbot for advice during a difficult emotional situation, suggesting millions of Americans may do so—mostly without harm. Still, the cases that reach court highlight the gap between what these tools can do and what they should do in a crisis.