Microsoft’s Mustafa Suleyman warns users not to confuse AI chatbots with real human beings: Here’s why

Mustafa Suleyman

Mustafa Suleyman, the CEO of Microsoft AI, has issued a clear and urgent warning: artificial intelligence may sound human, but it is not human — and treating it as such could have dangerous consequences.

In a personal blog post running over 4,600 words, Suleyman wrote about the growing trend of people ascribing human-like qualities to AI chatbots. Thanks to rapid advancements in generative AI, users are now turning to these tools not just for information but also as therapists, friends, and even romantic partners. Suleyman calls this phenomenon “seemingly conscious AI” — systems that may convince people they are a new kind of “person.”

He stresses that these views are his own and not Microsoft’s official stance, but they come from someone leading one of the biggest forces in AI innovation. And his concerns are serious.

“Human consciousness is hard to define,” Suleyman explains. But traits like memory, goal setting, problem-solving, and the ability to communicate with personality can trick us into believing an AI has a mind of its own. The risk, he says, is that we start granting machines human-like autonomy and rights, which could fuel psychological vulnerabilities, worsen dependence, and add a new layer of polarization in an already divided society.

Examples of AI’s misuse already exist. There have been troubling cases of chatbot “therapists” encouraging self-harm, and platforms like Character.AI have faced lawsuits over alleged links to teen suicides. Children and teenagers, in particular, are seen as most vulnerable to these dangers.

Instead of focusing on “model welfare,” Suleyman argues, society must prioritize protecting actual human beings, along with animals and the environment. He urges AI companies to explicitly clarify that their products are not conscious and to avoid encouraging users to believe otherwise. More research into how people interact with AI is also needed, along with greater transparency around safety measures and design principles.

Suleyman’s message is clear: AI may feel startlingly human, but mistaking it for the real thing could create more harm than good.