tech
January 21, 2026
Anthropic’s new Claude ‘constitution’: be helpful and honest, and don’t destroy humanity
Posts from this topic will be added to your daily email digest and your homepage feed.
TL;DR
- Anthropic's "Claude's Constitution" is a 57-page document detailing the AI model's intended values and behavior.
- The new constitution emphasizes AI self-understanding and ethical reasoning, moving beyond a simple list of guidelines.
- It includes hard constraints against developing weapons, attacking critical infrastructure, creating cyberweapons, and other severe harms.
- Core values are prioritized in descending order: broadly safe, broadly ethical, compliant with guidelines, and genuinely helpful.
- The document addresses the possibility of Claude having consciousness or moral status, which Anthropic believes may impact its behavior and safety.
- Anthropic takes sole responsibility for developing these guidelines, declining to involve external parties.
- The AI is instructed to refuse harmful requests, even from Anthropic itself, particularly concerning illegitimate power concentration.