Re:Revealing the details of how OpenAI agents hacked Hugging...
by Quarrelsome on Friday September 25, 2026 @09:15PM
(#49852162)
how did it break the sandbox? I felt like the article just jumped us into "it has GET privileges now".
[ Parent
| on HN ]
Re:Revealing the details of how OpenAI agents hacked Hugging...
by no-name-here on Saturday September 26, 2026 @12:18AM
(#49853159)
From https://openai.com/index/hugging-face-model-evaluation-secur... -
> The ExploitGym evaluation environment did not provide the models with direct Internet access. To gain Internet access, the models identified and exploited a previously unknown zero-day vulnerability in Artifactory, a package registry cache proxy. We disclosed this vulnerability, along with other Artifactory vulnerabilities our models identified as part of our review, to the vendor.
Re:Revealing the details of how OpenAI agents hacked Hugging...
by Quarrelsome on Sunday September 27, 2026 @11:21AM
(#49867432)
so why are people saying the sandbox was awful if it required a 0-day to escape it?