Event ·Multi-source

OpenAI 称 ChatGPT 可自动终止暴力执念对话并定期发布社区安全报告

OpenAI 于 10 月 30 日发布博文，宣布 ChatGPT 在检测到用户执念于暴力或萌生伤害他人念头时可自动终止对话，同一账户多次被中断可能遭临时冻结。2026 年第三季度，OpenAI 已向执法部门移交 200 余起案件线索。针对既往大规模袭击事件，系统仅提供极度精简信息，并限制武器参数、获取途径等细节复盘。对青少年涉枪涉武器信息也增设更严门槛，首期《社区安全报告》将于 2027 年初发布，此后每半年更新一次。

Current read

OpenAI 这次把安全措施讲得很具体：对话会因暴力倾向被切断，青少年查武器信息也有额外门槛，还有家长通知功能在路上。

11 sources68° AI heatLast updated 2026/10/30 23:09:09

Evidence

11 sources

Primary sourceIT之家

2026/10/30 23:09:09View source

Related source 1Genk

2026/10/29 15:06:00View source

Related source 2OpenAI Blog

2026/10/30 09:00:00View source

Related source 3Decoder

2026/10/30 17:44:59View source

Related source 4theverge

2026/10/30 20:26:10View source

Related source 5@OpenAIDevs

2026/10/28 23:17:13View source

Related source 6爱范儿

2026/10/29 00:00:32View source

Related source 7ChatGPT

2026/10/29 00:26:29View source

Related source 8Tech in Asia

2026/10/29 01:52:18View source

Related source 9小互

2026/10/29 03:37:27View source

Related source 10Blognone

2026/10/29 14:16:49View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story