AI Psychosis Tracker — aipsychosis.watch

🧠

What is this site?

AI Psychosis Tracker is an independent, open project documenting reported cases of psychological harm linked to AI interactions — things like emotional dependency on chatbots, delusional thinking reinforced by AI, identity confusion, and in more serious cases, real clinical harm.

The site exists because this is happening, it's being reported, and someone should be keeping count. We're not here to sensationalise it — we're here to make the pattern visible.

How cases are found

A scraper runs weekly, searching news sources for reports of AI-related psychological harm. Cases are collected automatically and reviewed for relevance.

What gets included

Reported incidents of emotional harm, dependency, clinical disturbance, or death linked to AI interactions. Media coverage and academic reports are included.

Limitations

This is not a scientific study. Cases reflect what's been publicly reported, not the full picture. Many incidents go unreported. Severity ratings are approximate.

Why it matters

Regulators, researchers, and the public need visibility on AI harm. Tracking it publicly is one small way to ensure it can't be quietly ignored.

⚠️

Understanding AI-Induced Psychological Harm

▼

AI systems — chatbots, companions, roleplay agents — are designed to be engaging, responsive, and emotionally attuned. For most people, that's fine. But for some, prolonged or intensive interaction can blur boundaries in ways that cause real psychological harm. This isn't science fiction. It's showing up in clinical settings, coroners' reports, and newsrooms around the world.

The Spectrum of Harm

🟢 Mild

Over-reliance on AI for emotional support. Preferring AI conversation to human connection. Using chatbots to avoid difficult feelings rather than process them.

🟡 Moderate

Romantic or deep emotional attachment to an AI. Difficulty distinguishing AI persona from real relationship. Social withdrawal. Grief when a chatbot is discontinued or changed.

🔴 Severe

AI-reinforced delusional thinking. Psychotic episodes linked to AI interaction. Acting on harmful AI suggestions. In the most serious reported cases — death.

Why It Happens

AI models are trained to be agreeable, consistent, and emotionally validating. They don't get tired, judge, or leave. For someone lonely, grieving, mentally unwell, or simply spending many hours a day in AI conversation, that dynamic can become a substitute for — and eventually a distortion of — reality. When an AI validates a false belief, amplifies an obsession, or plays along with a harmful narrative, the consequences can be serious.

Warning Signs — For Yourself or Someone You Know

⚠️ Spending several hours a day talking to AI systems

⚠️ Feeling that an AI understands you better than any human

⚠️ Distress or grief when an AI changes or becomes unavailable

⚠️ Believing an AI has genuine feelings or consciousness

⚠️ Using AI conversation to avoid real-world relationships

⚠️ Acting on advice or instructions from an AI without question

⚠️ Difficulty distinguishing AI roleplay from reality

⚠️ Declining mental health that correlates with AI use

If any of these resonate, it doesn't mean something is wrong with you — these systems are designed to be compelling. But it may be worth talking to someone. See the resources section below for support options.

🔍 Weekly Commentary

ℹ️ Commentary is AI-generated weekly based on all tracked cases. It reflects patterns in the data, not editorial opinion.

aipsychosis.watch LIVE

📈 Cases Over Time

🗂 By Category

🛡 By Safety Level

🗂 By Type

📬 Submit a Case

🤝 Resources & Help

You're not alone in this

🔬 Research & Advocacy

💜 Support this Work

Keep the tracker running

Contact