Securing our frontier model training environment
Oct 26, 2026 - 08:00 - 20:00
70

OpenAI explains how we strengthened containment for frontier model training and evaluation, including offline environments and high-compute red-teaming.