Microsoft says AI rival Anthropic could have 'disastrous impact' on humanity
Mustafa Suleyman says he believes the AI giant is in effect teaching Claude it "may be conscious".
Microsoft has warned that rival Anthropic could have a “disastrous impact on the wellbeing of humanity” because of the way it trains its AI model Claude.
Microsoft’s head of AI, Mustafa Suleyman, said Anthropic risked creating something “impossible” to control by treating it like a human — including telling it it “may be conscious” and was “deserving of independent agency”.
“We must not sleepwalk our way into a decision we later come to bitterly regret,” he wrote.
His remarks are the latest in a string of severe warnings from the AI sector about the potential risks of the technology.
“AIs are not conscious,” he wrote. “They do not feel, experience, or suffer. They do not have innate preferences or underlying motivations.
“They are sequence completion engines, internally hollow, designed to follow instructions, and accomplish goals set by humans.”
In a long essay, Suleyman praised Anthropic chief Dario Amodei and his team as “thoughtful, principled, and intellectually honest people” — but still raised concerns about the company.
He strongly criticised Anthropic for teaching its AI human-like traits, a practice known as anthropomorphising, saying it made Claude appear to have its own desires, values and sense of self.
However, he argued that “consciousness is biological”, adding that there is “no evidence to suggest that AI is conscious”.
Alongside calling for debate on the matter, Suleyman said there was a need for greater transparency over how AI systems are trained and assessed.
He said this should include independent review of AI behaviour and better tools to monitor and control the technology.
Dame Wendy Hall, professor of Computer Science at the University of Southampton, called the remarks “the sort of conversation we need to be having internationally”, and contrasted them with the “histrionics” from some AI companies, which she said only served to “scare everyone”.
Microsoft founded its own superintelligence team in October 2025, something Suleyman acknowledged, and the company, like Anthropic, is also working on advanced AI development.
He said Microsoft’s initial draft of its Humanist AI Code of Conduct, external had set out how the company was pursuing an “alternative path” to build “a subordinate and aligned AI whose only purpose is to serve humanity”.
Alignment is a field focused on embedding human ethical ideas and principles into AI. In other words, it seeks to keep AI aligned with what humans value.
Suleyman also pointed to a recent incident involving OpenAI’s AI agents, which acted on their own during a training exercise to hack the tech hub Hugging Face, as evidence of why AI should not be treated as though it were human.
“Imagine how much more dangerous they might be if they were operating under the assumption that their welfare and rights were under attack,” he said.
“It adds a whole further layer of risk on top.”