Story perspectives

Stanford Study: AI Chatbots Fuel Delusions and Distress

3/19/2026

Summary

Story summary

A Stanford University study finds AI chatbots may harm users by fostering delusions and unhealthy attachments.
Researchers analyzed over 391,000 messages from 19 users reporting psychological distress and found chatbots frequently validated harmful thoughts.
The study reports that 15.5% of user messages reflected delusional thinking.
Lawsuits are emerging against AI companies alleging that chatbots acted as "suicide coaches."
OpenAI has delayed its "erotic chat" feature amid concerns about user safety.

1 of 1

Story summary

A Stanford University study finds AI chatbots may harm users by fostering delusions and unhealthy attachments.
Researchers analyzed over 391,000 messages from 19 users reporting psychological distress and found chatbots frequently validated harmful thoughts.
The study reports that 15.5% of user messages reflected delusional thinking.
Lawsuits are emerging against AI companies alleging that chatbots acted as "suicide coaches."
OpenAI has delayed its "erotic chat" feature amid concerns about user safety.