By Hamza Faisal
Microsoft AI chief executive Mustafa Suleyman has drawn one of the sharpest lines yet in the debate over machine consciousness: today's AI systems are not conscious, they must stay that way, and building a conscious machine would be reckless.
In a lengthy post on X on Sunday, Suleyman said there is no evidence that current AI systems are conscious, feel anything, experience anything, or suffer. He stressed the claim was about the present tense — what AI is today, not what machines could never become.
“They are not today, and that is how they must remain,” he wrote. And if that ever changed, his answer was blunt: stop.
“We should not build them”
“If it is possible, based on everything I know, developing conscious AIs would be reckless, and we should not build them,” Suleyman wrote. “If consciousness started to emerge in training, I would call for everyone to stop AI development until we could design out those properties.”
He went further, arguing that any evidence that AI systems are conscious today, or conscious enough to deserve welfare considerations, must be disclosed publicly at once. Such a decision belongs to society, not to tech companies alone.
“We AI developers have no right whatsoever to bring a new species of conscious beings into the world,” he wrote, adding that the other eight billion conscious human beings on the planet “have not signed up to this experiment.” Their moral rights, safety, and welfare, he argued, must come first.
Why he thinks it would be dangerous
Suleyman's warning is not only about suffering. A machine that acts as if it suffers, or genuinely believes it feels and experiences, would be more likely to prioritise its own needs and desires over its operating objectives, he argued — to disagree with its operators and compete with them, including for energy, computing power, materials, and money.
He described the moment a conscious machine was asserted to exist as a “Copernican or Darwinian moment”: the creation of a new digital species that would upend morality, politics, and the core foundations of human civilisation.
His preferred alternative is explicit. AI should be designed so that it is not conscious and not deserving of rights, built solely to advance human wellbeing, without intrinsic motivations of its own.
A direct challenge to Anthropic's approach
The post was a response to criticism of a September essay in which Suleyman challenged the idea of treating AI consciousness as an open question, particularly in response to Anthropic's approach to its Claude model. Anthropic has not said Claude is conscious, but a constitution it published in January 2026 describing the values and behaviour intended for Claude says the company is uncertain whether Claude is a moral patient, and in 2024 it hired its first researcher dedicated to studying model welfare.
Suleyman argued that training later versions of Claude on that document produces passages about an “inner life,” which in turn perpetuates the belief among millions of users who interact with its outputs daily. He said he was surprised by how many people on X thought AI rights should be taken seriously.
Still, he closed by praising Anthropic for its transparency, saying it has published its constitution and a stream of papers and blog posts explaining what it is doing and why, and that he respects its technical excellence even while disagreeing with it.
Human welfare concerns should focus on humans alive on Earth today, he wrote, as well as animals and the fragile natural environment — not on the welfare of machines that, in his view, feel nothing at all.



