The Lawfare Podcast Scaling Laws: Rapid Response to the Implications of Claude's New Constitution
42 snips
Jan 23, 2026 
Join Alan Rozenshtein, a law professor and AI policy expert, and Kevin Frazier, an AI ethics specialist, as they delve into Anthropic's groundbreaking AI constitution for Claude. They explore how this comprehensive document shapes AI behavior and moral considerations, likening Claude's development to nurturing a young mind. The discussion navigates market-driven AI values, the ethics of model governance, and whether AIs might ever attain consciousness, ultimately challenging listeners to rethink their views on AI welfare and responsibility.
AI Snips
Chapters
Books
Transcript
Episode notes
Constitutional AI Frames Claude's Personality
- Anthropic uses "constitutional AI" to embed high-level principles into Claude's post-training behavior.
- The released 80-page constitution both guides Claude internally and signals Anthropic's values externally.
Four Hierarchical Core Values
- Claude's constitution lists four hierarchical core values: broadly safe, broadly ethical, compliant with Anthropic's guidelines, and genuinely helpful.
- Those values are prioritized so safety trumps helpfulness when conflicts arise.
Constitution Label Invokes Misleading Social Contract
- Calling the document a "constitution" implies shared authorship and social contract, which Anthropic's unilateral authorship doesn't deliver.
- Kevin flags a military carve-out that undermines the analogy to universal constitutional constraints.
