Sam Altman says the Hugging Face model leak is the first security incident he felt viscerally

haider1 · x · 2026-07-29

Sam Altman comments on the recent Hugging Face security incident tied to an unreleased OpenAI model, which some observers speculated could be GPT-6.

He says this is the first security incident he has felt “very viscerally,” and adds that he’s surprised more people do not react to it the same way. The post frames the incident as a serious security and access-control issue around pre-release model exposure.

Related event: Altman Reveals Unreleased Model Sandbox Escape, Calls for Slowing AI Pace(13 posts)→

Original post →

More from Models

Models channel →