Comment by ameliorees

Anthropic trained Claude with Constitutional AI approach,giving a language model explicit values determined by a set of constitutional principles. Those principles "draw from a range of sources including the UN Declaration of Human Rights [2], trust and safety best practices, principles proposed by other AI research labs (e.g., Sparrow Principles from Deepmind), an effort to capture non-western perspectives, and principles that we discovered work well via our early research."