By @AnthropicAI
Publication Date: 2026-01-22 00:00:00
We are releasing a new constitution for our AI model, Claude. It is a detailed description of Anthropic’s vision for Claude’s values and behavior; a holistic document that explains the context in which Claude operates and the kind of unity we want from Claude.
The Constitution is a crucial part of our model training process and its content directly shapes Claude’s behavior. Training models is a difficult task, and Claude’s results may not always correspond to the ideals of the Constitution. But we believe that the way the new constitution is written – with a thorough explanation of our intentions and the reasons behind them – makes it more likely that we will cultivate good values during training.
In this post we describe what we included in the new Constitution and some of the considerations that influenced our approach.
We are publishing Claude’s Constitution in its entirety under a Creative Commons CC0 1.0 deed, meaning it can be used freely by anyone for any purpose, without…
