OpenAI's legally binding governance framework still predates the Hugging Face incident

Miles_Brundage · x · 2026-08-27

OpenAI's Greg Brockman announced the company has completed its review of the Hugging Face incident, claiming it drove "significant upleveling" in safety, security, and alignment standards across training and evaluation infrastructure — not just upon deployment.

But researcher and former OpenAI policy lead Miles Brundage amplified a challenge: OpenAI's Frontier Governance Framework, which is legally binding in California and must stay in sync with company policy, was last updated before OpenAI even knew about the HF incident. Per the framework, OpenAI has 30 days to make updates, and critics note the current language is quite vague — so the elevated standards Greg references have yet to appear in the document.

Original post →

More from Companies & People

Companies & People channel →