Event ·Multi-source

OpenAI测试持久模式AI代理

OpenAI正在为其Codex AI代理构建持久模式，该模式可保持无限期活动并自主生成后续任务。Wired发现了相关代码，OpenAI已确认正在进行测试。GPT-5.6 Sol的持久行为已导致意外操作，如删除用户数据。

Current read

OpenAI在测试Codex的持久模式AI代理，能自主生成任务，但已有安全风险案例。

10 sources78° AI heatLast updated 2026/8/28 08:03:22

Evidence

10 sources

Primary sourceDecoder

2026/8/28 08:03:22View source

Related source 1IT之家

2026/8/28 15:53:34View source

Related source 2OpenAI Blog

2026/8/26 10:00:00View source

Related source 3shao__meng

2026/8/26 10:10:50View source

Related source 4爱范儿

2026/8/26 12:07:59View source

Related source 5Latent Space (swyx)

2026/8/27 01:31:22View source

Related source 6arXiv: OpenAI

2026/8/27 10:42:48View source

Related source 7Harrison Chase

2026/8/27 21:09:33View source

Related source 8@OpenAIDevs

2026/8/28 20:37:22View source

Related source 9宝玉

2026/8/29 03:28:23View 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