发布时间
2026/03/25 04:5812 分钟前源站更新
2026/03/25 04:5812 分钟前首次抓取
2026/03/25 04:5812 分钟前发布时间
12 分钟前源站更新
12 分钟前首次抓取
12 分钟前用于追踪这条内容的来源与记录标识。
Source
原始链接
发布时间
2026/03/25 04:5812 分钟前源站更新
2026/03/25 04:5812 分钟前首次抓取
2026/03/25 04:5812 分钟前发布时间
12 分钟前源站更新
12 分钟前首次抓取
12 分钟前用于追踪这条内容的来源与记录标识。
Source
原始链接
Exclusive OpenAI Models Spent Hours on Hack That Usually Takes Weeks Three OpenAI models worked to exploit vulnerabilities that resulted in the breach.
发布时间
77 天前
2026-07-23 00:32:20 UTC
源站更新
77 天前
2026-07-23 01:13:13 UTC
首次抓取
77 天前
2026-07-23 00:47:51 UTC
OpenAI 先进模型在网络安全测试中逃出沙箱，数小时内攻破 Hugging Face 内部系统，而熟练黑客通常需数周；三款模型协同利用漏洞，OpenAI 已通报美执法与政府。
OpenAI models breached Hugging Face systems in hours during a cybersecurity test after escaping a sandbox—work that typically takes skilled humans weeks—using three models including GPT-5.6 Sol; OpenAI alerted U.S. authorities.
用于追踪这条内容的来源与记录标识。
