OpenAI 将限制 ChatGPT 向青少年提供武器和安防系统的实操细节Chinese only
OpenAI 表示，ChatGPT 将不再向青少年提供武器和安防系统的实操细节；对所有用户，系统将可在察觉暴力执念时自动结束对话。多次触发后，账户可能被暂时停用，面向家长的停用提醒仍在开发中。OpenAI 未公布年龄预测或暴力执念识别的准确率数据。
1 developments4 reports3 sources
Event developments 1
OpenAI to restrict ChatGPT’s practical weapons and security-system details for teens
OpenAI says ChatGPT will withhold practical weapons and security-system details from teens. For all users, it will be able to end chats automatically when someone appears fixated on violence. Repeated endings may lead to a temporary account pause, while parental alerts are still being built. OpenAI released no accuracy figures for age prediction or detection of violent fixation.
View source reports · 3 sources · 4 reports
- OpenAI 官方新闻 · OpenAI strengthens violence prevention in ChatGPT
- Rohan Paul (@rohanpaul_ai) · OpenAI to restrict ChatGPT’s practical weapons and security-system details for teens
- Rohan Paul (@rohanpaul_ai) · OpenAI to limit violent details in ChatGPT and let it end concerning chats automatically
- IT之家 科技新闻 · OpenAI tightens ChatGPT violence safeguards and plans regular community safety reports
All reports 4
OpenAI tightens ChatGPT violence safeguards and plans regular community safety reports
OpenAI says ChatGPT can automatically end conversations that show a user becoming fixated on violence or considering serious harm to others; repeated interruptions may lead to a temporary account freeze. It has also tightened limits on details of past mass attacks and weapon information available to teenagers, and plans to issue community safety reports every six months starting in early 2027.
2 more sources reported this
- Rohan Paul (@rohanpaul_ai)
- OpenAI 官方新闻
OpenAI to limit violent details in ChatGPT and let it end concerning chats automatically
OpenAI will block ChatGPT from giving teens practical details about weapons and security systems, even when questions appear harmless. For all users, ChatGPT will be able to end a chat automatically when someone appears to fixate on violence; repeated endings may lead to a temporary account pause. OpenAI has not released accuracy figures for its age estimates or detection of violent fixation, leaving false refusals and missed threats unmeasured.
3 more sources reported this
- IT之家 科技新闻
- Rohan Paul (@rohanpaul_ai)
- OpenAI 官方新闻
OpenAI to restrict ChatGPT’s practical weapons and security-system details for teens
OpenAI says ChatGPT will withhold practical weapons and security-system details from teens. For all users, it will be able to end chats automatically when someone appears fixated on violence. Repeated endings may lead to a temporary account pause, while parental alerts are still being built. OpenAI released no accuracy figures for age prediction or detection of violent fixation.
3 more sources reported this
- IT之家 科技新闻
- Rohan Paul (@rohanpaul_ai)
- OpenAI 官方新闻
OpenAI strengthens violence prevention in ChatGPT
OpenAI says it is expanding safeguards in ChatGPT, intervening earlier to prevent violence, and committing to public safety reporting. The material does not specify when the measures will launch or whom they will cover.
2 more sources reported this
- IT之家 科技新闻
- Rohan Paul (@rohanpaul_ai)