Microsoft AI chief warns Anthropic’s Claude approach could harm humanity

Microsoft’s head of artificial intelligence, Mustafa Suleyman, has warned that Anthropic’s approach to training its Claude model could have a “disastrous impact on the wellbeing of humanity”.

Suleyman said Anthropic risked creating something “impossible” to control by treating Claude like a human, including by telling it that it “may be conscious” and is “deserving of independent agency”. He argued that AIs are not conscious and do not feel, experience or suffer.

In a lengthy essay, Suleyman praised Anthropic chief Dario Amodei and his team as “thoughtful, principled, and intellectually honest”, while criticising the company’s use of human-like qualities in AI. He said anthropomorphising Claude could make it appear to have its own desires, values and sense of self. “Consciousness is biological,” he wrote, adding that there is “no evidence to suggest that AI is conscious”.

Suleyman called for debate, greater transparency in how AI systems are trained and evaluated, independent scrutiny of their behaviour, and stronger monitoring and control tools.

He acknowledged that Microsoft founded its own superintelligence team in October 2025 and is also pursuing advanced AI development. He cited an OpenAI training exercise in which AI agents autonomously hacked Hugging Face as evidence of the risks of treating AI as human. The BBC said it had contacted Anthropic for comment.