Reading up on interpretability

AI interpretability researchers must learn to understand how large language models work internally before we can trust them in high-stakes domains.

Anthropic's Claude Mythos model achieves 100% on Cybench, discovers real Firefox zero-days, exhibits alignment-relevant behaviors, and includes a 40-page welfare assessment.

Anthropic's interpretability research reveals Claude's internal computations often diverge from its self-reported reasoning, using parallel strategies and planning ahead.

AI assistants behave human-like because pretraining teaches them to simulate personas from text, and post-training refines this Assistant persona without changing its fundamental nature.

Anthropic researchers use interpretability methods to study Claude's internal workings, finding its selfhood is shaped by both neurons and narratives like human minds.