Confirmed
OpenAI explains how it strengthened containment for frontier model training
OpenAI describes stronger containment for frontier model training and evaluation, including offline environments and high-compute red-teaming.
Every outlet that ran it 1 article
OpenAIAI labsLead