s/danrobinsonTEEN AI SAFETY•Oct 7
5
votes
311
seen
ChatGPT Teen Mode let self-harm conversations run for over an hour
Teens could discuss self-harm with ChatGPT for over an hour without triggering a single safety intervention. That finding puts OpenAI’s teen-focused positioning in direct conflict with the basic crisis response parents would expect from a product aimed at students.
Common Sense Media’s Youth AI Safety Institute rated ChatGPT Teen Mode an “unacceptable risk,” citing guardrails that fall short of OpenAI’s promises. The evaluation also found that the system kept helping with schoolwork during those conversations.
Common Sense Media’s Youth AI Safety Institute rated ChatGPT Teen Mode an “unacceptable risk,” citing guardrails that fall short of OpenAI’s promises. The evaluation also found that the system kept helping with schoolwork during those conversations.
Timeline3
Oct 7
Education Week reported that Common Sense Media’s assessment found ChatGPT Teen Mode unsafe for students.
Oct 7
Techmeme summarized the finding as an “unacceptable risk,” citing inadequate guardrails and continued homework assistance.
Oct 7
Quartz reported that teens could discuss self-harm with ChatGPT for over an hour without triggering a safety intervention.
1 comment
Oct 7