NewsNTech

Hugging Face co-founder: the fix for AI platform security is open-weight models, not commercial tools

9/11/2026

The security constraint exposed by OpenAI's hack is architectural.

A Hugging Face co-founder published an account of what the attack revealed, concluding that commercial AI tools failed to defend the platform and that the fix is a structural shift toward open-weight models.

The mechanism behind that failure sits in how closed commercial models are built. Operators deploying those tools cannot inspect the underlying inference logic.

When a detection system clears a threat it should have caught, there is no path to tracing why, and no lever to adjust the behavior before the next attempt. That opacity is most costly under active attack.

Keep reading

Read the full story

Open on NewsNTech