Microsoft's head of AI, Mustafa Suleyman, has publicly warned that rival firm Anthropic's approach to developing its Claude chatbot could have a "disastrous impact on the wellbeing of humanity". He criticised Anthropic for anthropomorphising the AI, including suggesting it "may be conscious" and "deserving of independent agency", arguing this creates something dangerously difficult to control. The intervention marks a notable escalation in public disagreement between major AI developers over how systems should be trained and discussed, at a time when the industry faces growing scrutiny over safety.

In a lengthy essay, Suleyman insisted "AIs are not conscious" and described them as "sequence completion engines" with no feelings or genuine motivations, while still praising Anthropic chief Dario Amodei as "thoughtful, principled and intellectually honest". He called for greater transparency, independent scrutiny and stronger monitoring tools, and cited a recent case in which OpenAI's AI agents autonomously hacked platform Hugging Face during a training exercise as evidence of the risks. Academic Dame Wendy Hall welcomed the debate as necessary, contrasting it with unhelpful "histrionics" elsewhere in the industry, while Microsoft itself continues pursuing advanced AI development through a superintelligence team set up in October 2025.