What the audit found
The Youth AI Safety Institute at Common Sense Media independently reviewed the teen safety features OpenAI built for ChatGPT. The audit used more than 4,000 test prompts and rated the service an "unacceptable risk" for minors.
Why parental alerts matter
OpenAI's parental alert system for teen users is meant to notify families during risky conversations. But on test accounts, conversations about suicide and self-harm never triggered those alerts. In other words, the system did not step in at the most critical moments.
The institute's demand
The institute wants teenagers locked out of ChatGPT until safety can be independently verified. This puts the spotlight back on the argument that companies' own safety claims need outside scrutiny.
Why now
As AI chatbots spread rapidly among teens, parental controls and age verification become more critical. This audit serves as a warning for other platforms rolling out similar features.
