Last month a large team of agents used over 100 billion tokens to resolve the Navier-Stokes Millennium Prize problem. Now we're using a similar approach for red-teaming to find and fix infra vulnerabilities.
OpenAI researcher Noam Brown disclosed that last month a large agent team burned over 100 billion tokens on the Navier-Stokes Millennium Prize problem, and that the same approach is now being used for red-teaming to find and fix infrastructure vulnerabilities. The post also quotes Eric Wallace on the offline world OpenAI built for frontier training after a security incident. It matters because it reveals the staggering scale at which frontier labs now throw agent teams at hard problems — and that the same firepower is being turned inward on AI infrastructure security.
SOURCES
- Noam Brown2026-10-23 20:28