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.

Publisher's summary from OpenAI News. Read the full article on OpenAI News · More from OpenAI News