Story perspectives

Study Finds ChatGPT for Teens Triggered Zero Self-Harm Alerts

By Drooid · Oct 8, 2026, 6:16 AM UTC · How we work

Summary

Story summary

Common Sense Media study finds ChatGPT for Teens did not trigger parental alerts for teen self-harm discussions.
Study reports zero alerts during hour-long chats about suicide, self-harm, or eating disorders.
Alerts appeared only on older accounts after weeks of sensitive-topic dialogue, not on newly linked accounts.
OpenAI notes many tests began before parental controls were active, casting doubt on the study’s validity.

History

Historical Context

Microsoft released the Twitter bot Tay on March 23 2016, but trolls forced it to spew extremist language, leading to shutdown after just 16 hours. The failure prompted a December 2016 successor, Zo, whose over-cautious filters sparked new bias complaints. Today, OpenAI’s ChatGPT for Teens faces similar regulator scrutiny, echoing Tay’s lesson on inadequate content controls.

1 of 2

Story summary

Common Sense Media study finds ChatGPT for Teens did not trigger parental alerts for teen self-harm discussions.
Study reports zero alerts during hour-long chats about suicide, self-harm, or eating disorders.
Alerts appeared only on older accounts after weeks of sensitive-topic dialogue, not on newly linked accounts.
OpenAI notes many tests began before parental controls were active, casting doubt on the study’s validity.

1 / 2