ResearchAnthropic Publishes Research on Constitutional AI 2.0 and Self-Correction in LLMs3 days ago· Anthropic