Anthropic's core values for Claude

Friday July 10, 2026

These are "Claude’s core values" from the "Constitution" for Claude that Anthropic published early in 2026. "In cases of apparent conflict, Claude should generally prioritize these properties in the order in which they’re listed."

One interesting comparison is to Asimov's laws of robotics. Law one on not harming humans is similar to safe/ethical, and law two on obeying orders is close to "genuinely helpful." Law three on protecting its own existence has at least a partial mapping to complying with company guidelines. (How idealistic of Asimov, to put the company behind the user in the priority order.)

Another interesting comparison is to Anthropic's 2023 Constitution for Claude, which was more of a laundry list of principles, drawing on the UN Universal Declaration of Human Rights, Apple’s Terms of Service, explicit "Consideration of Non-Western Perspectives," DeepMind’s Sparrow Rules, and two lists from Anthropic's research.

Anthropic articulates some of the differences between focusing on rules vs. values (see "Our approach" in the full constitution). This footnote from the release post has particularly clear phrasing:

Training on rigid rules might negatively affect a model’s character more generally. For example, imagine we trained Claude to follow a rule like “Always recommend professional help when discussing emotional topics.” This might be well-intentioned, but it could have unintended consequences: Claude might start modeling itself as an entity that cares more about bureaucratic box-ticking—always ensuring that a specific recommendation is made—rather than actually helping people.