Anthropic releases a new behavior charter for Claude, seeking to enable it to make its own moral judgments in the future

Anthropic announced the release of a new charter for its AI model Claude, which details Anthropic's vision for Claude's values and behaviors.

In the new charter, Anthropic upgrades Claude's soul document from a rulebook to a textbook that combines values and reasoning, hoping that Claude can become wiser, safer, and capable of making moral judgments on its own, rather than strictly adhering to rules. This is currently the most detailed publicly available AI personality setting and ethical constitution document.