IT之家 · 智能时代T2·Industry
58 pts
OpenAI has launched a CSAM image classifier, offered free via API to vetted organizations, which cuts false-flagged images by 48% and roughly doubles the likelihood that flagged content is genuinely abusive. ChatGPT adds an intervention flow that suspends accounts for 24 hours on related requests, escalating to permanent bans or law-enforcement referral. NCMEC supported training and evaluation, and OpenAI joined Anthropic and Google in commissioning a Tech Coalition guide on handling child sexual exploitation disclosures.