HHH Framework

A foundational AI alignment framework proposed by Anthropic, requiring models to simultaneously satisfy three dimensions: Helpful, Harmless, and Honest. It represents one of the core paradigms in mainstream AI alignment theory, guiding the design of safety-oriented language models.