OpenAI red-teamed their hardened containment infrastructure by asking teams of frontier-model agents to try to break through its defenses. They easily escaped in novel ways. After strengthening its containment architecture, OpenAI resumed its frontier RL training and inference on October 1. https://openai.com/index/securing-our-frontier-model-training-environment/ See also: Models are now approaching a massive general jump in cyber capabilities. From the Kimi K3.1 launch post: '...we found that without data removal at either the mid-training or the RL stage, downstream cyber capability would increase substantially.'

openai.comopenai.comKimi K3.1 Pro Tech Blog: Advancing Open IntelligenceIntroducing Kimi K3.1 Pro, our most capable 3T-class open-weight model with 1M-token context and major advances in coding, 3D modeling, and financial research.www.kimi.ai