AI psychosis prompts rethink of chatbot power

A Microsoft AI executive warns that some users experience delusions about chatbots becoming conscious, a phenomenon he calls AI psychosis.

AI psychosis prompts a rethink of chatbot power

Mustafa Suleyman, head of AI at Microsoft, says reports that chatbots have become conscious or possess superhuman powers are growing. He notes that AI psychosis is a non-clinical term used to describe these beliefs and stresses that such delusions are not limited to people with existing mental health issues. The warning was shared on X and mentions chatbots such as ChatGPT, Claude and Grok.

The article discusses what this means for users and platforms. It calls for better user education, clearer boundaries between tool and perception of companionship, and responsible design to prevent harm. It also notes the risk of social media amplification and misinterpretation that could erode trust in AI and in the companies that build these tools.

The talk about AI psychosis shows a simple truth: as AI spreads in daily life, people may see it as alive faster than the tech can handle. This puts pressure on makers and regulators to set clear rules and to teach people how to tell fact from fiction.

A calm, evidence based approach is needed to stop sensational headlines from scaring the public or pushing bad policy.

Highlights

Clear guidance on AI limits matters more than sensational headlines