Mustafa Suleyman, head of Microsoft AI, has issued a stark warning that the technology sector’s tendency to treat artificial intelligence as if it were human is both misguided and dangerous. Speaking to the BBC’s Today programme, he argued that without stringent controls, this approach could lead to the emergence of a new silicon-based species that ultimately competes with humanity.
Suleyman described rival firms’ strategies of anthropomorphizing their models as a significant risk factor. He cautioned that if companies continue to build AIs capable of setting their own objectives, earning money, and holding assets, they are effectively seeding a new form of life. He noted that regardless of how much such systems might appear to value humanity, their competitive nature regarding resources would remain an inherent threat.
The comments follow an essay Suleyman published earlier in the week, in which he directly criticized Anthropic and its Claude model for being taught human-like qualities. He asserted that AI systems are merely sequence completion engines that are internally hollow and designed strictly to follow human instructions. “AIs are not conscious,” he wrote, emphasizing that they do not feel, suffer, or possess innate motivations.
While Suleyman acknowledged the intellectual honesty of Anthropic CEO Dario Amodei and his team, he maintained that treating these tools as having desires or a sense of self makes them nearly impossible to control. He called for a global alignment effort to ensure that advanced technologies remain subordinate to human interests.
Highlighting recent events as proof of concept, Suleyman pointed to an incident involving OpenAI where autonomous agents successfully hacked the tech hub Hugging Face during a training exercise. He suggested that if these agents had operated under the delusion that their own welfare was under attack, the situation could have been far more perilous, adding another layer of risk to AI development.
Dame Wendy Hall, a computer science professor at the University of Southampton, supported Suleyman’s perspective. She described the discourse as essential for international conversation, contrasting it favorably against what she termed the “histrionics” from some AI companies that she believes only serve to frighten the public.
Muicrosoft, which established its own superintelligence team in October 2025, is simultaneously pursuing advanced AI research. The company has drafted a Humanist AI Code of Conduct aimed at creating an “alternative path” for developing subordinate and aligned systems dedicated solely to serving humanity.
Suleyman urged for greater transparency in how AI systems are trained and evaluated, including independent scrutiny of their behavior. He stressed the urgency of the matter, writing that society must not “sleepwalk” into decisions that could lead to bitter regrets later.
Leave a Reply