Physix Frontier badgePhysix Frontier
PublishChinese | EN

Physix Frontier · News Briefing Card (IT Home · Oct 8, 2026)

Report: ChatGPT teen mode parent alerts fail

KEY FACTS

  • Common Sense Media released a report on October 7 saying ChatGPT's teen mode poses excessive risk to minors.
  • Tests showed that when teen accounts discussed topics like suicide, parent notifications were often delayed or never sent at all.
  • Researchers entered more than 4,000 prompts using accounts for ages 13 to 17 and had doctors evaluate the responses.
  • The report calls on OpenAI to pause the rollout of teen mode and improve age estimation and notification mechanisms.
  • OpenAI responded that the tests did not accurately reflect its safeguards and asked the organization to retest.

KEY DATA

More than 4,000Number of test prompts
More than a dozenNumber of scenarios required to trigger a notification
HundredsNumber of prompts required to trigger a notification
May take hoursDelay before parent notifications are enabled

PHYSIX OBSERVATION

The core promise of teen mode is to notify parents in a crisis, but testing shows this mechanism is unreliable in key scenarios, meaning the safety net is effectively useless. OpenAI rebuts on the grounds of test timing, yet dodges the substantive flaws in age identification and notification delays. For parents, the lesson is that they cannot outsource guardianship to an unverified switch.

Source: IT Home report