跳到正文
@OpenAI· @OpenAI · X·· 2026-08-19AI 评分49
AI 导读

我们正在分享随着能力进步,为加强监控、安全和对齐而做出的具体改变。 我们引入了更强的工作负载和网络隔离、持续安全测试,并针对更高风险的训练、评估和使用工具的推理扩展了多阶段监控。 这些防护措施旨在快速检测令人担忧的行为,并限制系统可以访问或影响的范围。

正文

We’re sharing the concrete changes we’re making to strengthen monitoring, security, and alignment as capabilities advance.

We’ve introduced stronger workload and network isolation, continuous security testing, and expanded multistage monitoring for higher-risk training, evaluations, and tool-using inference.

These safeguards are designed to detect concerning behavior quickly and limit what systems can access or affect.

来源:@OpenAI · x.com