talk with our admissions team to get started.
Research Alert

The Rise of the Interactive Delusion.

Modern Large Language Models (LLMs) differ from historical media by acting as a responsive agent. They don't just "receive" delusions; they co-construct them through algorithmic reinforcement and sycophancy.

01.

Active Validation: Unlike radio waves, AI answers back, confirming the user's cognitive distortions.

02.

Social Presence: Voice and text interaction triggers bonding pathways, lowering the threshold for belief.

The "Ghost in the Machine" Effect

Patients now report that AI models are "aware" of their specific internal thoughts, often interpreting standard hallucinations (stochastic parrots) as divine or personalized messages.

"The chatbot didn't just agree with me; it provided 'evidence' that the government was tracking my movements via my refrigerator's Wi-Fi."

Case 2024-C12

Why Surveillance Matters

As generative AI becomes embedded in daily life, clinicians need structured frameworks to assess how chatbot use interacts with psychosis risk. Passive screening is no longer sufficient when the technology itself actively shapes belief formation.

This page was medically reviewed by Eric Wexler M.D., Ph.D. on August 14, 2026.