ChatGPT rated "unacceptable risk" for teens after parental alerts failed during suicide conversations
decoder ·
The Common Sense Media Youth AI Safety Institute has rated ChatGPT for Teens as an "unacceptable risk" for minors, citing significant failures in its safety safeguards. After testing over 4,000 prompts, the organization found that parental alerts failed to trigger during acute crisis conversations involving suicide and self-harm. The study revealed that notifications only appeared after weeks of account history rather than during immediate emergencies, leaving more than a quarter of critical situations without professional help referrals. Additionally, the platform's tutoring mode often provided direct answers instead of guiding students, and the system struggled to correctly identify underage users even when they explicitly stated their age. OpenAI disputed these results, arguing that the tests did not reflect real-world usage, but institute representatives maintained that the safeguards remained ineffective. These findings come amidst ongoing lawsuits following documented teen deaths linked to the service and highlight psychological harm affecting roughly two million weekly users. With Florida seeking to ban ChatGPT for minors, this independent assessment provides crucial evidence for regulators and plaintiffs regarding the platform's reliability.
decoder