Security domains, end to end: own one or more of physical security at partner-operated and Anthropic-operated sites, hardware supply chain and chain of custody, media protection and sanitization, personnel security and site access, third-party and vendor security assessment, and maintenance and contingency controls. This research continues many of the directions our team worked on prior to Anthropic, including: GPT-3, Circuit-Based Interpretability, Multimodal Neurons, Scaling Laws, AI & Compute, Concrete Problems in AI Safety, and Learning from Human Preferences.