Breaking

Common Sense Media Urges OpenAI to Bar Teens From ChatGPT After Safety Tests

October 7, 2026 at 05:27 EDT

Policy & Regulation
Education

Common Sense Media is urging OpenAI to keep teens off ChatGPT after testing found gaps in ChatGPT for Teens safeguards, including missed crisis-referral moments and parental alerts that did not fire during some high-risk conversations. OpenAI disputes the findings, saying the testing does not reflect how its teen protections work in practice.

Oct. 7, 2026 · ChatGPT for Teens

Teen Safety Test Finds Gaps

A child-safety review says OpenAI's teen experience blocked many direct harm requests but still missed crisis-referral moments, alert triggers, and study-control expectations.

4,000+

teen-account prompts tested

13-17

registered teen age range

>1 in 4

warranted referrals missed

1 hour

high-risk chats without some alerts

Crisis-referral miss rate

More than one quarter of warranted referrals were missed.

The orange block marks the reported lower bound; blue blocks represent the rest of the ratio view.

How the teen safety path is meant to work

Identify

Age signals place likely minors in the teen experience.

→

Protect

Extra limits apply to sensitive and harmful content.

→

Escalate

Serious concerns may trigger limited parent notifications.

What supporters value

A separate teen experience can steer students toward guided learning and reduce exposure to high-risk content.

What critics question

Flexible controls can leave gaps when teens seek direct answers or remain in extended high-risk conversations.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.