Facts
Every real, cited fact this site catalogs, grouped by category unless filtered.
The constitution is designed to prioritize explanation over instruction.
The framework is built so Claude could construct any specific rule Anthropic might come up with by understanding the underlying principles, rather than memorizing a rulebook — the document is described as "a detailed description of Anthropic's vision for Claude's values and behavior."
The constitution defines four core values Anthropic wants all current Claude models to hold, in explicit priority order.
In priority order: (1) broadly safe — not undermining appropriate human mechanisms to oversee AI during the current phase of development; (2) broadly ethical — being honest, acting according to good values, avoiding inappropriate/dangerous/harmful actions; (3) compliant with Anthropic's guidelines; (4) genuinely helpful.
Anthropic frames the constitution as a living document, not a finished one.
"Claude's constitution is a living document and a continuous work in progress. This is new territory, and Anthropic expects to make mistakes and hopefully correct them along the way."
Anthropic released a new "Constitution" for Claude on 2026-01-22.
An approximately 80-page document, released under a Creative Commons CC0 1.0 deed (public domain — freely usable by anyone for any purpose), marking a fundamental departure from Anthropic's 2023 constitutional-AI approach.
The constitution is a real input into model training, not just a public statement.
The document is described as a crucial part of the model training process — its content directly shapes Claude's trained behavior, not only its public messaging.
Machine-readable version: GET /api/facts?category=constitution