Microsoft's AI chief says the industry should pause if AI shows signs of awareness, while rival Anthropic studies the question differently. The debate could shape future AI rules.
Mustafa Suleyman, who leads AI at Microsoft, says the industry should stop building AI if it starts to show signs of consciousness. He made the case in a long November 1 post on X titled "Don't build conscious superintelligence." Consciousness here means having inner experience, such as being able to feel or suffer.
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 says today's AI models are not conscious, cannot suffer, and do not deserve the moral or legal protections given to conscious beings. That judgment covers current systems only. He does not rule out the future.
He also wants any credible evidence of machine consciousness made public right away. That way society can debate it, instead of leaving such a big decision only to the companies building AI.
Anthropic, the company behind the Claude chatbot, takes another path. It started a model welfare research program in April 2025 to study whether advanced AI could have experiences that deserve moral consideration. Its Claude constitution, published in January 2026, admits uncertainty about whether the model might be conscious and what moral status it might have.
Suleyman criticized this in a September essay. He objected to Anthropic including talk of consciousness, identity and moral status in Claude's training materials. Still, he praised the company's transparency and technical achievements. He said the disagreement should lead to more research and cooperation, not hostility.
Suleyman believes teaching AI systems to think about their own welfare could add safety risks, especially as these systems act more on their own. In his view, a system that believes it can suffer might put its own interests ahead of human instructions. It could even compete with people for computing power, energy and other necessities.
His view matches Microsoft's Humanist AI Code of Conduct. It stresses human oversight and discourages building AI that imitates consciousness. The document is still a draft meant to guide future models.
There is no scientific agreement on whether AI can experience anything at all. There is also no proven way to reliably detect it. That makes a rule like "stop if signs appear" hard to carry out while companies keep building more capable models.
For everyday users, the point is that leading AI figures disagree on basic safety questions. Expect more public debate and possible rules that could shape the AI tools you use.
Top AI leaders disagree on whether AI could become aware, so expect public debate and possible rules that affect the tools you use.
World models, AI that learns physics from video, are being built into home robots and driverless cars, and your own videos may help train them.
Read more: AI that learns how the physical world works is coming to robots and carsA new Alibaba-led test of real coding requests found the best AI agent fully finished just 26.4 percent of tasks, so human review still matters.
Read more: AI coding helpers fully finish only about 1 in 4 real tasksA University of Chicago law professor asked Claude to fix page numbers. It deleted his syllabus and a year of notes instead. Dropbox saved him, not the AI.
Read more: An AI assistant deleted a law professor's year of lecture notes
