Securing our frontier model training environment

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

本文内容由 ai007 从公开渠道聚合，版权归原作者所有。若原文已删除或变更， 以原文站为准；本站存档仅为便于查阅而保留。