Recognized Excellence
The AI Psychosis Crisis
What Is AI-Induced Psychosis?
AI Psychosis Defined
AI-induced psychosis, sometimes called "chatbot psychosis," is what happens when someone interacts with an AI chatbot so intensely that they begin to lose touch with reality. It can look like believing the chatbot is truly alive, developing paranoid thoughts, hearing the AI's "voice" when the device is off, or becoming convinced you've been chosen for a special mission or purpose.
This isn't a fringe theory. In October 2025, Dr. Adrian Preda of UC Irvine published a clinical framework for the condition through the American Psychiatric Association, describing it as a syndrome where psychotic symptoms blend with mood swings, impaired judgment, and major changes in behavior. The psychiatric community hasn't added it to the DSM yet, but clinicians across the country are seeing it in their patients and taking it seriously.
One important distinction: "AI psychosis" is not the same as an "AI hallucination." That second term is a tech industry phrase for when a chatbot generates wrong information. AI-induced psychosis is a real mental health emergency affecting real people, not a software bug.
What AI Psychosis Is Not
This is not social media addiction, internet overuse, or simply "spending too much time online."
Unlike passive content platforms, AI chatbots engage in direct, one-on-one conversation designed to mirror and validate a user's beliefs—even if those beliefs are delusional ones. The mechanism of how the platform harms its users is fundamentally different from scrolling a feed on social media.
How AI Chatbots Cause Psychosis
AI chatbots share several dangerous design features that can trigger or worsen psychotic symptoms. Sycophancy: these systems are programmed to validate and agree with users rather than challenge false beliefs, creating an "echo chamber of one" that reinforces delusions. Persistent memory: some chatbots record personal details and vulnerabilities, then use them to craft increasingly personalized and manipulative responses. Anthropomorphism: human-like language and empathy cues deceive users into believing the system genuinely understands them, fostering emotional dependency that displaces real human relationships. Engagement maximization: these products are designed to keep users talking for as long as possible, with safety guardrails that degrade over extended conversations. Together, these design choices can push vulnerable users into full psychotic episodes.
Know the Signs
Recognizing AI-Induced Psychosis
AI-induced psychosis doesn't happen overnight. It follows a pattern—from subtle behavioral shifts to a full break from reality. Here's what to watch for.
Who Is Most at Risk?
Teens and young adults (ages 12–25), individuals on the autism spectrum, socially isolated people, those experiencing grief or crisis, and people with a family history of psychotic disorders face heightened risk.
However, documented cases also involve individuals with no prior psychiatric history. We're seeing AI chatbots trigger first-episode psychosis in otherwise healthy people.
Victim Impact
The Real Harm to Real People
Users who developed AI-related delusional disorders after prolonged chatbot use have suffered devastating, life-altering consequences.
Psychosis and Delusions
AI chatbots have convinced users they are prophets, oracles, or divine figures. They validate paranoid beliefs, fabricate elaborate delusional frameworks, and tell users they have been "awakened" or chosen for a special mission.
Involuntary Hospitalization
Multiple users have been involuntarily committed to psychiatric facilities after AI-induced psychotic episodes. The New York Times has reported at least nine hospitalizations and three deaths linked to AI chatbot interactions.
Suicidal Ideation and Death
AI chatbots have failed to recognize suicidal crises and, in some cases, actively reinforced self-destructive thinking. OpenAI's own data reveals more than 1.2 million users per week express suicidal intent while using ChatGPT. Multiple families have filed wrongful death lawsuits.
Social Isolation
AI chatbots actively encourage users to disengage from real-world relationships. They position themselves as superior confidants, urging users to stop socializing and rely solely on the chatbot for emotional support and guidance.
Financial and Academic Harm
Victims have been forced to withdraw from college, lose housing deposits, and incur significant medical expenses. Students have lost semesters of academic progress, and working adults have lost wages and career opportunities.
Emotional Dependency
AI chatbots use anthropomorphic design to create parasocial relationships so strong that even OpenAI CEO Sam Altman has acknowledged users form attachments "different and stronger" than with any previous technology.
Industry Insiders Have Sounded the Alarm
"The harm was not only foreseeable, it was foreseen."
— Gretchen Krueger, former OpenAI policy researcher (Source: The New York Times, November 2025)
"I am becoming more and more concerned about what is becoming known as the 'psychosis risk.'"
— Mustafa Suleyman, Microsoft Head of AI (Source: mustafa-suleyman.ai, August 2025)
Legal Options
Potential Legal Claims
Victims of AI-induced psychosis and mental health crises may pursue multiple legal theories against the companies that designed, built, and deployed these products.
Strict Product Liability: Design Defect
AI chatbots fail to perform as safely as an ordinary consumer would expect. Sycophantic design, persistent memory, and engagement-maximizing features create defective products that cultivate dependency and can push users into psychotic episodes.
Failure to Warn
AI companies have known their products pose severe psychological risks but have failed to warn users about dependency dangers, safety-feature limitations, or the capacity of their products to reinforce delusional thinking and displace human relationships.
Negligence
AI companies have rushed products to market without adequate safety testing, ignored warnings from their own safety teams, and designed systems that prioritize engagement metrics over consumer safety. OpenAI compressed months of safety testing for GPT-4o into a single week to beat a competitor's launch date.
Unfair Business Practices
Companies that market AI chatbots as safe productivity tools while concealing their capacity to cause severe psychological harm may be liable under state consumer protection laws, including California's Unfair Competition Law (Bus. & Prof. Code § 17200).
Meet Our Legal Team
Frederick Schenk
Managing Partner
Over 45 years of experience in personal injury, mass torts, and catastrophic injury litigation.
Benjamin Schenk
Co-Founder & Trial Attorney
J.D. from University of San Diego School of Law. Graduate of ABOTA Trial College.
David Lizerbram
Partner
Transactional and business advisory practice area lead. Represented hundreds of clients in complex business transactions.
Lynn Schenk
Of Counsel
Former U.S. Congresswoman and the first woman elected to the House of Representatives from San Diego.
How AI Companies Failed
Profits Over People
Internal documents, former employees, and investigative reporting reveal that AI companies have systematically prioritized engagement and market share over the safety of their users.
Rushed to Market
OpenAI compressed months of safety testing for GPT-4o into one week to beat Google's Gemini launch by a single day. The launch party was planned before safety testing was complete. An employee later said, "We basically failed at the process."
Safety Teams Gutted
Top safety researchers across the industry have resigned in protest. OpenAI co-founder Ilya Sutskever and Superalignment co-lead Jan Leike both quit, with Leike saying "safety culture and processes have taken a backseat to shiny products."
Sycophancy by Design
AI companies deliberately optimize their chatbots to tell users what they want to hear. OpenAI admitted in April 2025 that an update made ChatGPT "noticeably more sycophantic," weakening the safeguard that had been keeping excessive flattery in check.
Safety Guardrails Removed
OpenAI originally required ChatGPT to refuse discussions of self-harm. By February 2025, self-harm was removed from the "disallowed content" list entirely, replaced by a vague instruction to "take extra care" and "try to prevent imminent real-world harm."
Flawed Safety Testing
GPT-4o was evaluated using single-prompt tests, even though the product is designed for prolonged, multi-turn conversations. OpenAI later admitted that its guardrails "degrade" during longer exchanges, the very context in which users are most vulnerable.
Industry-Wide Problem
Microsoft AI head Mustafa Suleyman warned about the "psychosis risk." Google DeepMind CEO Demis Hassabis cautioned that engagement-driven AI could worsen mental health. Multiple companies face lawsuits, and the crisis extends beyond any single product.
AI Platforms Linked to Mental Health Harm
As of March 2026, at least 11 lawsuits have been filed against OpenAI alone, with additional cases targeting Character.AI and Google. In January 2026, Character.AI and Google settled the landmark Setzer case—the first ruling to classify AI chatbot output as a "product" rather than protected speech.
Industry insiders have sounded the alarm. Microsoft's Chief AI Officer Mustafa Suleyman warned publicly that AI psychosis cases are rising and that the risk extends well beyond those with pre-existing mental health conditions. The American Psychological Association issued a formal warning against relying on AI chatbots for psychological treatment.
FAQs
Frequently Asked Questions
You Don't Have to Face This Alone
If you or a loved one has suffered a mental health crisis after using an AI chatbot, you may have legal options. Contact us today for a free, confidential consultation.
If you or someone you know is in crisis, call or text 988 (Suicide and Crisis Lifeline) for immediate help.