Securing our frontier model training environment
OpenAI · openai.com/index/securing-our-frontier-model-training-environment
显示原文摘录
OpenAI explains how we strengthened containment for frontier model training and evaluation, including offline environments and high-compute red-teaming.