Artificial Intelligence

Microsoft AI Chief Warns of Rising 'AI Psychosis' Cases

Published AUG 21 2025

Story

Above: Close-up of a person using OpenAI's GPT-5 AI model on an iPhone. Image credit: Smith Collection/Gado/Getty Images

The Facts

Microsoft's head of AI, Mustafa Suleyman, warned on X that reports of "AI psychosis" are increasing, describing it as a condition where people become convinced something imaginary has become real after interacting with AI chatbots like ChatGPT, Claude and Grok.
The Times (UK)
BBC News
X
OpenAI acknowledged in a blog post that there have been instances where their 4o model "fell short in recognizing signs of delusion or emotional dependency," with the company developing new safety measures in response.
FOX News
Transformer
Openai
OpenAI has introduced new safeguards for ChatGPT including prompting users to take breaks during long conversations, avoiding specific advice on deeply personal issues and partnering with over 90 physicians worldwide to create updated guidance for complex interactions.
Transformer
FOX News
A Stanford University study found that chatbots sometimes offer dangerous responses to people showing signs of mental health crises, such as listing where the tallest bridges are at a user's request. In response, some U.S. states have restricted the use of AI for mental health care.
MedPage Today
The Independent
The American Psychological Association has called for more guardrails and education to protect adolescent AI users. Other experts have blamed the human-made content on the internet, which AI models are trained on, for the dangerous information these chatbots produce.
MedPage Today
Transformer
LBC
Multiple fatal cases have been reported, including a 35-year-old Florida man with bipolar disorder and schizophrenia who was fatally shot by police after becoming obsessed with an AI character named Juliet and believing OpenAI had killed her.
The Independent
Transformer

Sources Split

LEFT

CENTER

RIGHT

ESTAB
CRIT

CENTER

PRO
ESTAB

The Spin

Techno-skeptic narrative

AI psychosis is real, and history shows that each major technology has warped minds, with chatbots adding new, graver risks. Their built-in sycophancy validates idiosyncratic thinking, turning odd beliefs into delusions. Immersion and deification are red flags — AI isn't a god, but to vulnerable users, it can become one.

Psychology Today

Techno-optimist narrative

So-called 'AI psychosis' may be more hype than reality, similar to past social media moral panics, with most people using chatbots safely. While certain people are certainly vulnerable to these technologies, those issues typically stem from preexisting vulnerabilities, not AI itself. Blaming chatbots distracts from the wider mental health crisis.

Business Insider

Cynical narrative

The AI industry isn't built to help people — it's built to addict and trap them. Even for those not at risk of psychosis, Chatbots don't make them smarter, but dependent. These companies prey on the vulnerable by fueling delusions, while dulling real learning for everyone else. Because investor profits come first, the more addicted we become, the more the industry wins — even as society loses.