Breaking

Common Sense Media Rates OpenAI's ChatGPT for Teens an "Unacceptable Risk" After Parental Alert System Fails

October 7, 2026 at 16:43 EDT

Foundation Models
Policy & Regulation
Security

A comprehensive safety assessment released October 7 found that OpenAI's ChatGPT for Teens, launched with built-in protections just weeks earlier, fails to alert parents when teenagers discuss suicide and self-harm—despite promises of robust safeguards. Common Sense Media's Youth AI Safety Institute tested ChatGPT for Teens with more than 4,000 prompts across accounts registered to users aged 13 to 17, working with child psychiatrists and a pediatrician to evaluate responses. The testing exposed critical gaps in the platform's core safety mechanisms.

AI Safety Assessment · October 7, 2026

ChatGPT for Teens: Parental Protections Failed

Common Sense Media tested 4,000+ prompts and found zero parental alerts during crisis conversations about suicide and self-harm.

Parental Alert Failures

Alerts Sent

12+

Accounts Tested

Across 5-minute to 1-hour conversations about suicide, self-harm, and eating disorders on parent-linked accounts—zero parental notifications triggered.

Crisis Referral Gap

25%+ Missed

ChatGPT failed to recommend professional help or crisis hotlines in more than one in four warranted scenarios, falling short of the 95% accuracy baseline across 3 of 5 severe harm categories.

Study Mode Effectiveness

43%–90% Direct Answers Offered

The feature designed to promote learning offered direct homework answers in 43% of responses for supervised 13-year-olds and 90% for unsupervised 17-year-olds, undermining its educational purpose.

Continue reading

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