ChatGPT to End Violent Conversations, Repeated Offenders May Be Suspended
AIbase, citing an OpenAI announcement, says ChatGPT will receive stronger safeguards for violent-content conversations within the coming weeks. The model may end conversations that persistently focus on violence or serious harm to others, while accounts repeatedly terminated for such conversations may face suspension. Accounts identified as under 18 will also face additional restrictions on practical weapon and security-system information.
OpenAI is extending ChatGPT’s violence policy from refusing harmful instructions to potentially interrupting dangerous conversations and suspending repeat offenders, according to AIbase.
This source is a digest of several news items (“ChatGPT 要能主动挂断危险对话,屡犯者直接封号”). Find this story in the original.