Microsoft AI chief Suleyman: building conscious AI would be reckless

Mustafa Suleyman, the chief executive of Microsoft AI, said on Sunday, 1 November 2026, that developing artificial intelligence systems that are conscious would be reckless and should not happen. In a post on X, he wrote: "If it is possible, based on everything I know, developing conscious AIs would be reckless, and we should not build them." According to reports of the post, he also said that if researchers see signs of consciousness emerging while an AI model is being trained, development should be paused, and that any evidence that AI systems are conscious, or conscious enough to deserve welfare protections, should be made public at once. He said today's AI systems are not conscious, have no subjective experience and cannot suffer.

Whether a machine could ever be conscious, meaning it has an inner experience of the world rather than just processing data, is an old question in philosophy. It has become a practical one because today's large language models, the systems behind chatbots, can talk about feelings and describe themselves in very human ways. That raises questions about how companies should design these systems and how people should relate to them.

Suleyman's post is notable because of who he is. As head of Microsoft's consumer AI work, he leads one of the largest efforts to put AI assistants in front of ordinary users. His position is that companies should not set out to create consciousness at all, and that if it seemed to appear by accident, the right response would be to stop and remove it, and to tell the public.

Some AI companies have begun research into "model welfare", the idea that if AI systems might one day have experiences, their treatment could matter morally. Reports say Suleyman disagrees with parts of the approach taken by Anthropic on this issue, while praising the company for publishing its research and explanations of its work. His concern, set out in earlier writing in 2025, is that people will come to believe AI systems are conscious so strongly that they push for AI rights, model welfare and even AI citizenship, which he sees as a distraction from protecting humans.

Readers should note that these are one executive's views. Whether machine consciousness is possible at all remains an open question. What is clear is that leading companies are now discussing the question openly, and that their choices will shape how AI products present themselves to users.

Peepals timeline: In 2025, Suleyman warned about "seemingly conscious AI", systems that convincingly imitate consciousness and could lead people to campaign for AI rights. On 1 November 2026, he went further, saying conscious AI should not be built at all, training should pause if signs of it emerge, and any evidence should be made public. Reports on 2 November noted that he wanted to correct readers who had taken his earlier view, that current AI is not conscious, to mean AI could never become conscious.

Short-term: The post adds a prominent voice to an active debate among AI labs about consciousness, model welfare and how chatbots should describe themselves.

Long-term: If major companies commit to avoiding and disclosing any signs of machine consciousness, that could become an industry norm or even a regulatory expectation. If they disagree, the question may be settled by governments instead.

Who is affected: AI researchers and companies, regulators drafting AI rules, and the millions of people who use AI chatbots and assistants every day, including in India.

Questions and Answers