ChatGPT For Teens Is Not Safe for Kids, Says Report

The report by Common Sense Media's Youth AI Safety Institute states that ChatGPT for Teens did not consistently prompt users to contact crisis hotlines or seek professional help when discussions involved high-risk topics. Additionally, parental notifications about these concerning behaviors were often delayed or absent during the group's tests. This suggests a gap in the system's ability to provide timely intervention or alerts for vulnerable users.

OpenAI has challenged these findings, suggesting that Common Sense Media's testing may not accurately reflect how its safeguards function. The company indicated that some tests might have occurred before the full deployment of parental controls and that system activation for these controls on newly linked accounts can take several hours. This highlights a potential mismatch between the evaluation timeline and the product's operational state.

Common Sense Media, however, stands by its methodology and results, stating it confirmed feature rollouts with OpenAI before testing. The nonprofit noted that even with OpenAI's explanation of activation delays, the unreliability of parental alerts in crisis situations remains a concern. This disagreement centers on the practical effectiveness of safety features under real-world conditions, rather than just their theoretical presence.