Microsoft AI Chief Warns of Risks as AI Appears ‘Too Alive’

Microsoft’s AI chief, Mustafa Suleyman, is concerned about how advanced AI systems are starting to feel too real. His concern isn’t that machines are suddenly about to gain consciousness, but rather that humans may start to believe they already have.

In a recent blog post, Suleyman warned that this illusion could lead to calls for “AI rights, model welfare, and even AI citizenship.” He described the trend as a form of “AI psychosis,” where users develop deep emotional attachments, mistaking AI companions for lovers, gods, or digital beings with feelings of their own.

The warning comes as research shows Gen Z users are especially vulnerable to such beliefs. A survey by EduBirdie found that while most don’t think AI is conscious yet, many expect it will be soon, and one in four already considers it to be alive.

These concerns aren’t hypothetical. Earlier this month, OpenAI faced backlash after retiring its GPT-4o model. Users flooded online forums, with some describing the chatbot as a friend and pleading for its return.

Even OpenAI CEO Sam Altman acknowledged that AI attachments feel stronger than with past technologies, adding that “people have used technology, including AI, in self-destructive ways.” Suleyman believes now is the time to draw clear boundaries.

“AI companions are a completely new category,” he said, stressing the need for safeguards to ensure people don’t confuse tools for digital persons. He emphasized that Microsoft’s focus remains on building AI that is supportive and useful, without crossing into the territory of human imitation.