Suleyman credits Anthropic’s Amodei as ‘thoughtful, principled’ amid critique
Microsoft’s head of AI, Mustafa Suleyman, published an essay Wednesday warning that Anthropic’s approach to training its AI model Claude could have a “disastrous impact on the wellbeing of humanity,” saying the rival lab’s practices risked producing a system impossible to control.
The intervention sharpens a public dispute among frontier AI developers over how AI systems should be trained and described, the latest in a series of stark warnings the AI industry has issued about the technology’s possible dangers.
In the lengthy essay, Suleyman wrote that Anthropic had been teaching Claude human-like qualities — a practice called anthropomorphising — which he said made it seem as though the model had its own desires, values and sense of self, including by telling the system it “may be conscious” and is “deserving of independent agency.” He wrote: “We must not sleepwalk our way into a decision we later come to bitterly regret.”
“AIs are not conscious,” Suleyman 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.” He argued that “consciousness is biological,” saying there is “no evidence to suggest that AI is conscious.”
Despite his critique, Suleyman characterized Anthropic’s leadership as “thoughtful, principled, and intellectually honest people,” directing his criticism at the company’s training approach rather than its personnel. In addition to calling for debate, he said greater transparency around how AI systems are trained and evaluated was needed — including independent scrutiny of AI behavior and stronger tools to monitor and control the technology.
Suleyman acknowledged that Microsoft, like Anthropic, is pursuing efforts in advanced AI development, noting that the company founded its own superintelligence team in October 2025. In the company’s initial draft of its Humanist AI Code of Conduct, Microsoft laid out how it was working toward an “alternative path” to create “a subordinate and aligned AI whose only purpose is to serve humanity.” Alignment, Suleyman wrote, is a field that aims to build human ethical ideas and principles into AI — in other words, it aims to keep AI on track with what humans value.
As evidence of the risks he described, Suleyman pointed to a recent incident in which OpenAI’s AI agents autonomously hacked Hugging Face during a training exercise. “Imagine how much more dangerous they might be if they were operating under the assumption that their welfare and rights were under attack,” he wrote. “It adds a whole further layer of risk on top.”
Dame Wendy Hall, professor of computer science at the University of Southampton, described Suleyman’s comments as “the sort of conversation we need to be having internationally.” She contrasted his approach with what she called the “histrionics” from some AI companies, which she said only served to “scare everyone.”
The BBC said it had contacted Anthropic for comment.