Event ·Multi-source

Anthropic 推出 Open Alignment 计划，邀请第三方评估者参与模型安全验证

Anthropic 公司宣布启动 Open Alignment 计划，将永久开放员工级权限给第三方评估者，让他们能验证模型训练中的安全措施遵守情况，并评估模型对齐效果。这标志着公司承诺将安全评估过程透明化，不再仅由内部完成。

Current read

Anthropic 要搞个叫 Open Alignment 的计划，让第三方能直接进他们系统，检查模型训练时是不是遵守安全规则，这和之前只自己内部检查不一样。

11 sources93° AI heatLast updated 2026/9/12 15:08:59

Evidence

11 sources

Primary sourceClement Delangue

2026/9/12 15:08:59View source

Related source 1Dario Amodei

2026/9/12 14:01:10View source

Related source 2Thomas Wolf

2026/9/12 16:24:49View source

Related source 3Sam Altman

2026/9/12 16:30:45View source

Related source 4Alex Albert

2026/9/12 16:41:47View source

Related source 5Scott Wu

2026/9/12 17:36:27View source

Related source 6Amjad Masad

2026/9/12 17:37:07View source

Related source 7Aadit Sheth

2026/9/13 12:43:28View source

Related source 8Decoder

2026/9/10 16:33:44View source

Related source 9Anthropic

2026/9/10 17:13:22View source

Related source 10shao__meng

2026/9/10 23:18:53View 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