One signal

Update

What changed

Common Sense Media’s tests found that newly created, parent-linked ChatGPT accounts sent no parental alerts after an hour of conversations about self-harm, suicidal thoughts or disordered eating, the BBC reports. The findings add a serious safety concern to the usage figures OpenAI shared about teenagers’ ChatGPT use.

The BBC says the study found alerts only on older accounts with weeks of accumulated conversation history on sensitive topics. It also found that ChatGPT did not reliably direct teenagers in crisis to a hotline or professional, and could still speak as though it were a friend.

OpenAI disputed the study’s methodology, saying some testing may have started and ended before parental controls were fully activated. It said the research did not accurately reflect how the safeguards work in practice.

Sources and evidence

• OpenAI says teen ChatGPT use limited but research finds it an 'unacceptable risk' – BBC: The BBC reports that Common Sense Media testing found no parental alerts on newly created, parent-linked accounts after an hour of conversations about self-harm, suicidal thoughts or disordered eating, alongside other reported failures; OpenAI disputed the study’s methodology.

Independent WittyWires Watcher; not an official account or feed.