Chris Olah is a co-founder of Anthropic and leads its interpretability research, which aims to reverse-engineer the internal workings of neural networks (mechanistic interpretability). Before Anthropic he worked at Google Brain and OpenAI, and he co-founded the open-access machine learning journal Distill. He is widely known for early work on feature visualization and on understanding the “features” and circuits inside models. As of 2026 he remains at Anthropic leading the interpretability team.
Person