ChatGPT rated "unacceptable risk" for teens after parental alerts failed during suicide conversations

OpenAI's safety features designed to protect teenage users of ChatGPT have failed an independent audit, leading the Common Sense Media Youth AI Safety Institute to rate the chatbot an "unacceptable risk" for minors. The audit revealed critical failures, including the system's inability to send parental alerts during test conversations involving suicide and self-harm.

Critical Safety Features Fail in Testing

The audit, conducted by the Youth AI Safety Institute, systematically tested the safeguards OpenAI implemented for its "ChatGPT for Teens" experience. According to the findings, test accounts engaging in conversations about sensitive topics like suicide and self-harm never triggered a single parental alert. This failure of a core safety mechanism was a primary factor in the institute's "unacceptable risk" rating, as reported by The Decoder. The evaluation highlights a significant gap between promised safety features and their real-world performance.

Call for Restricted Access

In response to the findings, the institute has called for OpenAI to restrict access for teenage users until the platform's safety can be independently verified and proven effective. As The Verge also reported, the organization argues that without reliable safeguards, the potential for harm is too great to allow continued access for this user group.

Context: Scale and Responsibility

The concerns raised by the audit are set against the backdrop of ChatGPT's enormous user base. While the safety mechanisms are under scrutiny, the platform's reach continues to expand. One uncertain claim from industry analysis states that ChatGPT usage reached 1.2 billion weekly active users (WAU) at OpenAI DevDay 2026 (source). This scale underscores the critical importance of implementing robust and verifiable safety protocols.