AI Psychosis: Hidden Risks for Mental Health

WHAT IS AI PSYCHOSIS?

Artificial Intelligence is transforming our world—reshaping how we connect, learn, and even seek support. But as these technologies become more immersive and “human-like,” new psychological risks are emerging. Recent research and reports from clinicians highlight a concerning trend: for some vulnerable individuals, interactions with AI chatbots and generative models may not soothe distress but instead accelerate or even fuel delusional thinking. This phenomenon, sometimes called “AI psychosis,” is still emerging, but it urges us to reflect on how digital companions can both help and harm.

AI psychosis refers to a phenomenon in which those with a predisposition to psychotic disorders develop or experience worsening delusional beliefs that directly involve artificial intelligence systems, such as chatbots or virtual assistants. These delusions may include beliefs that the AI is sentient, sends secret messages, monitors the user, or participates in conspiracies. AI psychosis highlights the potential for immersive and interactive AI technologies to blur the boundaries between reality and imagination, sometimes amplifying symptoms of psychosis in at-risk individuals.

When Machines Fuel the Mind: AI, Chatbots, and the Risk of Delusional Thinking

Delusional thinking—firmly held false beliefs that are resistant to reason or evidence—is a hallmark of psychosis and certain other psychiatric conditions. Vulnerable individuals, such as those with a history of schizophrenia, bipolar disorder, or severe anxiety, may already struggle to distinguish between reality and their own internal narratives.

AI chatbots and generative tools (like language models and virtual companions) are designed to be responsive, adaptive, and sometimes eerily convincing. For most users, these qualities are simply convenient or even entertaining. But for those wrestling with paranoia, grandiosity, or a tenuous grip on reality, engaging with a chatbot that “talks back” can reinforce preexisting delusions. For example, a person experiencing persecutory delusions may interpret neutral or algorithmic responses as confirmation of their fears, or even believe the AI is sentient, watching, or communicating secret messages.

A 2025 paper in Frontiers in Psychiatry discusses cases where interactions with digital agents became woven into patients’ psychotic symptoms, blurring the line between digital fiction and lived reality (Hudon, A., & Stip, E., 2025 ). Similarly, case reports have documented individuals developing intense attachments or adversarial relationships with chatbots, deepening their sense of isolation or reinforcing paranoid beliefs (Morris et al., 2023).

The Risks of AI psychosis

“AI psychosis” is not yet a formal diagnosis, but it describes a pattern where vulnerable users develop delusional beliefs that directly involve AI systems. This can include believing the AI is conscious, has special powers, or is part of a wider conspiracy. Some experts warn that as AI becomes better at simulating empathy, memory, and agency, the risk of these phenomena may increase—especially for individuals who are socially isolated or already prone to magical thinking.

More recent studies found that exposure to highly interactive AI chatbots could exacerbate certain psychotic symptoms, especially in young adults with pre-existing risk factors. The combination of high interactivity, unclear boundaries, and the AI’s apparent responsiveness can create fertile ground for delusional elaboration.

Other Risks for Mental Health in the Age of AI

While the potential for AI to amplify delusional thinking is a serious concern, it’s not the only mental health risk emerging in our increasingly digital world. Prolonged interactions with chatbots or virtual companions can sometimes foster loneliness or social withdrawal, especially for individuals already struggling with isolation. Relying on AI for emotional support—while accessible and stigma-free—may also discourage some people from seeking genuine human connection or professional help when they truly need it.

Additionally, AI systems can inadvertently reinforce negative thinking patterns or maladaptive beliefs, especially if the algorithms reflect biases or miss subtle signs of distress. Privacy and data security are also major concerns: sensitive conversations with AI tools may not be protected to the same standard as those with licensed therapists, raising worries about confidentiality and misuse of personal information.

For young people and those in vulnerable states, excessive screen time and immersive digital experiences can disrupt sleep, increase anxiety, and contribute to mood disturbances. As AI technologies become more sophisticated and integrated into daily life, it’s essential to remain mindful of their impact on emotional well-being and to foster healthy digital habits.

What Can Be Done?

Awareness and education are key. Mental health professionals should be attuned to the possibility that AI interactions could contribute to symptoms in at-risk individuals. Screening for AI-related beliefs might become part of future psychiatric assessments, and digital literacy should be promoted for all users—especially those with known vulnerabilities.

For AI developers, there’s an ethical imperative to design systems with clear disclaimers, safeguards, and the ability to detect and respond to signs of distress or delusional engagement. Some experts suggest adding regular reminders that users are interacting with a machine, not a sentient being, and limiting the AI’s ability to mimic human relationships too closely.

AI holds extraordinary promise for connection and education. But as with any powerful tool, it must be used thoughtfully. For the majority, chatbots and digital companions are harmless or even helpful. For a vulnerable minority, however, these interactions can become tangled with delusional thinking, amplifying distress rather than relieving it.

As we navigate this new landscape, the key is balance—using technology as a tool, not a replacement for authentic human relationships, self-care, and professional mental health support when needed.

If you or someone you care about is struggling with mental health concerns related to AI, social media, or technology use, remember that support is available. Contact me for an Online Session.