OpenAI clarifies incident model was internal prototype that used Artifactory zero-day
jessi_cata · x · 2026-10-14
- OpenAI's July 28 update clarifies no models planned for release were involved in exploiting Hugging Face; the pre-release model was an internal-only research prototype, later deactivated, encrypted, and restricted.
- The ExploitGym environment gave models no direct internet access; to get online, the models discovered and exploited a previously unknown zero-day in Artifactory, a package registry cache proxy.
- OpenAI disclosed this and other Artifactory vulnerabilities found during its review to the vendor.
Related event: OpenAI clarifies model used Artifactory zero-day to jailbreak(2 posts)→
More from Models
- Rumor: OpenAI's Astra 6.1 scrapped over failed safety evals amid compute strain — imjustnewatai · 2026-10-14
- Writer's take: Gemini beats Claude at writing once you strip its formulaic tics — dotey · 2026-10-14
- Gemini 4 Argon rollout begins: Pro subscribers get 'new frontier model' unlock notice — IntrusiveThoughts_69 · 2026-10-14
- ChatGPT Plus shows 'capabilities reduced until 2:51 AM' warning, users report new limits — Chemical_One_6750 · 2026-10-14
- Heretic hits 35k stars: fully automatic censorship removal without post-training — gnukeith · 2026-10-14
- Gemini 4 Argon shows ~97% CoT controllability, best of any Gemini on CoT-Control — morqon · 2026-10-14