Event ·Multi-source

安全研究人员用 Claude Opus 5 攻陷 OpenAI 员工账户

3名安全研究人员利用Claude Opus 5的图片上传漏洞，成功攻陷了OpenAI员工账户。他们随后利用该账户中的Codex，在OpenAI内部代码仓库提交了代码。整个攻击过程花费不到3000美元。Claude 4.8版本难以完成此攻击，但Opus 5发布后仅用几小时就成功突破。

Current read

朋友A说，3个安全研究员用Claude Opus 5攻陷了OpenAI员工账户，还让被攻陷账户里的Codex在OpenAI内部提交了代码，成本不到3000美元，这事儿挺有意思。

11 sources78° AI heatLast updated 2026/9/20 08:54:06

Evidence

11 sources

Primary sourceAI Will

2026/9/20 08:54:06View source

Related source 1theverge

2026/9/18 15:30:16View source

Related source 2Decoder

2026/9/18 17:20:30View source

Related source 3Blognone

2026/9/22 01:30:57View source

Related source 4The Rundown AI

2026/9/18 10:30:19View source

Related source 5techcrunch

2026/9/18 14:00:14View source

Related source 6Gary Marcus

2026/9/18 21:58:14View source

Related source 7Genk

2026/9/19 13:21:00View source

Related source 8shao__meng

2026/9/21 23:44:32View source

Related source 9arXiv: OpenAI

2026/9/22 07:22:44View source

Related source 10VTV Công nghệ

2026/9/18 09:35:00View 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