One doc tagged with "Interpretability"

What Anthropic's interpretability research says about hidden model reasoning — and what it changes for anyone operating agents.