A deep dive into Anthropic's role in AI safety, evaluating their practices and contributions across various categories.