Ex-OpenAI Policy Chief Pushes Frontier AI Audits

Episode 1 · Ex-OpenAI policy chief Brundage pushes for mandatory frontier AI audits (2026-09-16, 5 posts)

Miles Brundage, former head of policy at OpenAI, announced on September 16 that after leaving the company he will devote himself fully to "frontier AI auditing." He believes third-party audits are worth pursuing voluntarily across the industry, while stressing that audits are only one necessary condition—and are easy to render a box-checking exercise.

Confirmed

• Brundage laid out concrete regulatory proposals: auditing should become mandatory for leading AI companies, with minimum standards for auditor qualifications, independence, and access; real safety standards are needed rather than deeming compliance as "having a (possibly weak) policy and following it."
• He also called for policy directions such as mandatory incident disclosure and transparency.
• Addressing criticism over why he doesn't advocate pausing or slowing AI, Brundage responded: only after audits are made rigorous (high standards, heavy penalties for non-compliance, and mechanisms that make violations obvious) could a pause or slowdown realistically take hold; other governance work needs doing regardless, so there's no need to choose between a single action or framework. Once solid audit mechanisms are in place, necessary slowing of AI will happen naturally.
• He said he currently feels "okay" because the industry's general lack of adequate control over frontier AI is now being more widely recognized—but this is only a start: "there's a lot to do and not much time."

Why it matters

• Having held core policy and safety roles at OpenAI, Brundage's pivot to independently advancing audits represents a direct response from a former top-lab executive to the inadequacy of existing governance mechanisms, and could shape industry and regulatory agendas.
• He positions auditing as a precondition for more aggressive governance tools like pauses or slowdowns, offering an actionable middle path in the debate over "how to govern frontier AI."

• Ex-OpenAI Policy Head Miles Brundage Goes All-In on Frontier AI Auditing — Miles_Brundage · 2026-09-16
• Brundage: Mandate Audits at Leading AI Labs With Real Safety Standards — Miles_Brundage · 2026-09-16
• Ex-OpenAI safety lead Miles Brundage pushes AI incident disclosure and transparency rules — Miles_Brundage · 2026-09-16
• Brundage: nailing audits is a prerequisite for any pause to work well — Miles_Brundage · 2026-09-16
• Miles Brundage: Nail AI audits and any needed slowdown takes care of itself — Miles_Brundage · 2026-09-16

Episode 2 · Brundage Calls for Severe Penalties and Whistleblower Protections in AI (2026-09-16, 2 posts)

Brundage urged major investments in biosecurity and cyber defense, severe penalties for violations, strong whistleblower protections against NDA constraints, and funding for alignment research and defenses against nation-state AI-enabled attackers.

• Brundage Urges Whistleblower Protections and State-Level AI Security — Miles_Brundage · 2026-09-16
• Brundage: Serious Penalties Needed, Not Fines Below One Engineer's Salary — Miles_Brundage · 2026-09-16