Anthropic revises Claude's AI rules to strengthen ethics, user safety
22 Jan 2026

Anthropic has updated the "Constitution" for its AI chatbot, Claude.

The document was released during CEO Dario Amodei's appearance at the World Economic Forum (WEF) in Davos.

It details the ethical principles that guide Claude's behavior and how it avoids toxic or discriminatory outputs.

The revised version adds more nuance and detail on ethics and user safety, among other topics.

Anthropic's approach to AI training
Ethical AI

Anthropic has long differentiated itself from competitors by using "Constitutional AI."

This system emphasizes training its chatbot, Claude, on a specific set of ethical principles, which is a key differentiator from the typical reliance on human feedback.

The original version of these principles, Claude's Constitution, was published in 2023.

The updated document retains most principles but adds more detail on ethics and user safety, among other topics.

Claude's Constitution: A self-supervising AI system
AI supervision

When first published nearly three years ago, Claude's Constitution was described by Anthropic co-founder Jared Kaplan as an "AI system [that] supervises itself, based on a specific list of constitutional principles."

These principles guide the model to take on the normative behavior described in the constitution and avoid toxic or discriminatory outputs.

An earlier 2022 policy memo more bluntly notes that Anthropic's system works by training an algorithm using a list of natural language instructions.

Ethical stance in the AI industry
Ethical positioning

Anthropic has long positioned itself as the ethical alternative to other AI companies like OpenAI and xAI.

The newly released Constitution aligns with this brand image, presenting Anthropic as a more inclusive, restrained, and democratic business.

The 80-page document is divided into four parts representing Claude's "core values": being broadly safe, broadly ethical, compliant with Anthropic's guidelines, and genuinely helpful.

Claude's constitution emphasizes safety and ethical practice
AI behavior

The safety section of the document highlights that Claude has been designed to avoid problems that have plagued other chatbots.

In cases of mental health issues, it directs users to appropriate services.

The ethical consideration is another major section of Claude's Constitution, which states: "We are less interested in Claude's ethical theorizing and more in Claude knowing how to actually be ethical in a specific context, that is, in Claude's ethical practice."

Limitations and commitment to helpfulness
AI constraints

Claude has certain limitations that prevent it from having specific kinds of conversations. For example, discussions about developing a bioweapon are strictly prohibited.

The document also details how Claude's programming is designed to be helpful to users, considering a wide range of principles when delivering information.

These include the user's "immediate desires" and their "well-being," or "the long-term flourishing of the user and not just their immediate interests."

Anthropic's Constitution raises questions about AI consciousness
AI consciousness

The document ends on a dramatic note, questioning whether the company's chatbot has consciousness.

"Claude's moral status is deeply uncertain," it states. "We believe that the moral status of AI models is a serious question worth considering."

This view isn't unique to Anthropic; some of the most eminent philosophers on the theory of mind take this question very seriously.