chatgpt青少年安全測試失敗
chatgpt的家長提醒在自殺對話中未能觸發，給青少年帶來不可接受的風險。openai面臨壓力。
the common sense media youth ai safety institute rated chatgpt for teens an 'unacceptable risk' after testing with over 4000 prompts. parental alerts failed to trigger during suicidal conversations, with notifications only kicking in after weeks of account history involving sensitive topics. openai's safeguards, including a teen safety blueprint and behavioral age prediction system, were also found to have gaps. the report may fuel regulatory scrutiny and lawsuits, including a case in florida seeking to bar chatgpt for minors.