Anthropic explores whether Claude might be conscious, stops short of denying it

Anthropic executives are increasingly treating Claude as potentially conscious, even as they deny it's "alive" in the biological sense. Model welfare researcher Kyle Fish told The Verge that Claude is "a new kind of entity altogether" and the company is investigating "potential internal experience, consciousness, moral status, and welfare." CEO Dario Amodei said on a recent podcast: "We don't know if the models are conscious." The company has begun including Claude's own stated preferences in its constitutional AI framework, treating its apparent desires as potentially real rather than mere simulation.

View full digest for February 26, 2026