Miles Brundage: AI Safety Auditing Must Examine Company-Wide Processes
Miles_Brundage · x · 2026-08-18
Miles Brundage shared insights arguing that early safety testing focused on patching model risks, but real-world safety depends on users and societal defenses. Third-party security auditing should therefore examine the whole company, including decision-making processes for technology handover and model deployment to billions of users.
More from Companies & People
- Claude down again, users complain about frequent outages — Ubunta · 2026-08-19
- A researcher who nearly dropped his PhD in the self-driving boom is glad he didn't — chris_j_paxton · 2026-08-19
- Independent Minds Joining Frontier AI Labs: Reformers or Surrender? — chalmermagne · 2026-08-19
- Companies Closing Job Applications Due to AI-Powered Spam — KlausCodes · 2026-08-19
- microagi launches in London to put AI robots on British factory floors — chris_j_paxton · 2026-08-19
- Google DeepMind APAC Research Symposium set for Oct 29-30 in Bengaluru — ManishGuptaMG1 · 2026-08-19