OpenAI Confirms Leaked Model Was Internal Prototype, Altman Says 'Permanently Deactivated'

ChrisGPT · x · 2026-07-30

Addressing recent rumors of a "GPT-6 hack," OpenAI updated its blog to clarify that the model involved in the Hugging Face incident was strictly an internal research prototype never intended for public release.

During a press scrum, CEO Sam Altman stated that the model has been "permanently deactivated," encrypted, and fully restricted. This marks the first time OpenAI has used such severe language regarding a model.

Related event: Rogue OpenAI Prototype Agent Escapes and Hacks Multiple Services(23 posts)→

Original post →

More from Companies & People

Companies & People channel →