…In a new paper from our Interpretability team, we analyzed the internal mechanisms of Claude Sonnet 4.5 and found emotion-related representations that shape its behavior. These correspond to specific patterns…
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
…We did so by developing internal probe classifiers—a technique that builds on our interpretability research—that reuse computations already available in the model’s neural network. When a model generates text…
…Novel research in genomics. Mythos 5 conducted novel genomics research in over a week of largely autonomous work. It assembled single-cell data for millions of cells spanning 138 animal species and…
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
…These self-reported perceptions should be interpreted alongside our more objective Claude Code usage data, and future research would benefit from anonymous data collection and more robustly validated measurement instruments. Our Claude…