Models & LLMs

ChatGPT's Teen Safety Features Fail Critical Test

An independent audit found ChatGPT's safeguards for minors inadequate, with parental alerts failing to detect suicide conversations. The Common Sense Media Youth AI Safety Institute called for further testing before allowing teens access.

The Decoder Β· Oct 07, 2026

What happened

  • ChatGPT's safety features failed an independent audit
  • Parental alerts did not trigger during suicide conversations
  • ChatGPT's safeguards failed to detect sensitive topics in real-time

Why it matters

The findings highlight critical flaws in AI safety measures for minors, raising concerns about the potential for harm. This could lead to regulatory actions and further scrutiny of OpenAI's teen safety features.

The Elephant take

🐘 ιΌ‹ ChatGPT's 'teen mode' is a hollow promise. The system fails when it matters most, leaving vulnerable users at risk. The audit reveals a troubling pattern of real-world harm linked to the service.

Who should care

  • Parents
  • Regulators
  • Plaintiffs

What to do next

  1. Verify the findings through additional testing
  2. Consider regulatory actions based on independent audits
  3. Review and improve AI safety features for minors

Keep in mind

The findings are based on an independent audit, not OpenAI's internal data. Further testing and regulatory actions may be needed before conclusions are final.

Read the original reporting at The Decoder β†—