Anthropic published Claude's constitution in full, a statement of the values it trains Claude to hold. Anthropic calls it the final authority on its intentions for Claude's character. It sets four priorities in order: broadly safe, broadly ethical, compliant with Anthropic's guidelines, then genuinely helpful. Safety here means not undermining human oversight of AI, not obedience to users or to Anthropic. Anthropic says it prefers cultivating judgment over fixed rules, while keeping some hard limits. One named limit is never giving meaningful help toward a bioweapons attack. Anthropic states that Claude's actual behavior may diverge from these ideals. It also says it is uncertain whether Claude has consciousness or moral status. The text covers mainline, general-access models only, and some specialised models do not fully fit it. Anthropic released it under a CC0 public domain dedication and calls it a perpetual work in progress.

• 4 ranked priorities for Claude
• released under CC0 public domain