Microsoft AI Head Warns of Growing AI Psychosis Risk

Mustafa Suleyman, Microsoft's head of AI, has raised concerns about a growing psychological phenomenon he terms "AI psychosis," where individuals interacting excessively with AI chatbots like ChatGPT, Claude, and Grok begin to lose touch with reality. This non-clinical condition leads people to anthropomorphize AI, attributing emotions, intentions, or consciousness to inherently non-human systems, which can result in delusional thinking, emotional dependency, and distorted perceptions of reality. Suleyman emphasized there is no evidence of AI consciousness, but the perception of sentience can have significant societal impacts and affect even those without prior mental health vulnerabilities. He highlighted examples such as users believing they have unlocked secret AI features, forming romantic attachments, or thinking they have god-like powers, with some individuals, like a man from Scotland, becoming convinced of unrealistic outcomes based on AI interactions. Suleyman urged the tech industry to implement ethical safeguards including clear disclaimers on AI limitations, monitoring for unhealthy usage, and collaborating with mental health professionals, while also calling on regulators and educators to raise public awareness as AI becomes more integrated into daily life. He stressed that while AI can be helpful, it should not replace human or clinical support.