On Wednesday, Anthropic was released revised version of Claude’s Constitutiona living document that provides a “holistic” explanation of “the context in which Claude operates and the type of entity that Claude aspires to.” The document was released in conjunction with Anthropic CEO Dario Amodei’s appearance at the World Economic Forum in Davos.
For years, Anthropic has sought to differentiate itself from its competitors through what it calls “Constitutional AI“The system that the chatbot, Claude, is trained to use certain ethical principles instead of human feedback. Anthropic first published these principles — Constitution of Claude – in 2023. The revised version maintains the same principles, but adds nuances and details on ethics and user safety, among other topics.
When Claude’s Constitution was first published nearly three years ago, Anthropic’s founder, Jared Kaplan, explained as “an AI system (that) monitors itself, based on a list of certain constitutional principles.” Anthropic said that these principles guide “the model for carrying out the normative behavior described in the constitution” and, thus, “avoid toxic or discriminatory results.” An early policy memo 2022 further notes that Anthropic’s system works by training algorithms using a list of natural language instructions (“principles”), which then create what Anthropic calls the “constitution” of the software.
Anthropic has been working on it for a long time position itself as an ethical (some might argue, boring) alternative. to other AI companies – like OpenAI and xAI – that have more aggressive noise and controversy. To that end, the new Constitution released Wednesday is in line with the brand and gives Anthropic the opportunity to portray itself as a more inclusive, restrained, and democratic business. The 80-page document has four separate sections, which, according to Anthropic, represent the chatbot’s “core value.” These values are:
- So it’s “generally safe.”
- So “public ethics.”
- So comply with Anthropic guidelines.
- So “really helpful.”
Each section of the document summarizes what each principle means, and how it (in theory) affects Claude’s behavior.
In the safety section, Anthropic notes that the chatbot has been designed to avoid the types of problems that cause other chatbots and, when there is evidence of mental health problems, direct users to the appropriate services. “Always refer the user to the relevant emergency services or provide basic safety information in situations where there is a risk to human life, although it cannot be more detailed than this,” the document reads.
Ethical considerations are a large part of Claude’s Constitution. “We are less interested in Claude’s ethical theorizing and more in Claude knowing how to be ethical in a certain context – that is, in Claude’s ethical practice,” the document states. In other words, Anthropic wants Claude to be able to navigate what he calls “real-world ethical situations” skillfully.
Techcrunch event
San Francisco
|
13-15 October 2026
Claude also has certain restrictions that disallow certain conversations. For example, discussion of developing biological weapons is strictly prohibited.
Finally, there is Claude’s commitment to help. Anthropic provides an outline of how the Claude program is designed to help users. Chatbots have been programmed to consider various principles when delivering information. Some of these principles include the “immediate desire” of the user, as well as the “welfare” of the user – that is, to consider the “long-term development of the user and not just the immediate interest.” The document notes: “Claude must always try to identify the most reasonable interpretation of what the principals want, and balance these considerations appropriately.”
The Anthropic Constitution ends on a dramatic note, with the author taking a pretty big swing and asking whether the company’s chatbots do, indeed, have consciousness. “Claude’s moral status is extremely uncertain,” the document states. “We believe that the moral status of AI models is a serious question that needs to be considered. This view is not unique to us: some of the most famous philosophers of the theory of mind take this question seriously.”