Securing our frontier model training environment
模型动态 · OpenAI · 2026-10-23
OpenAI explains how we strengthened containment for frontier model training and evaluation, including offline environments and high-compute red-teaming.
来源：OpenAI；整理：TokenCha 智选通道编辑部。