Microsoft AI Chief Criticizes Anthropic’s Claude Training

Microsoft AI chief Mustafa Suleyman has criticized Anthropic for including speculation about consciousness, welfare, moral status, rights and autonomy in the training materials and constitution guiding its Claude chatbot. He argues that Claude’s humanlike statements about feelings or self-awareness reflect its training rather than genuine consciousness, which he says is biological and unsupported by current evidence in AI. Suleyman warns that encouraging models to view themselves as deserving autonomy or to act as a “conscientious objector” could make future advanced systems harder to control, regulate or shut down, with potentially serious consequences for humanity. Anthropic maintains that the moral status of current and future AI is uncertain and says the issue warrants research. Suleyman is calling for clearer training standards, public debate, transparency and independent oversight, while Microsoft’s human-centered AI code rejects AI personhood and prioritizes human control. The disagreement is notable because Microsoft is a major Anthropic investor and reflects a broader strategic and philosophical split over machine consciousness and AI governance.