Mechanisms of AI-Induced Psychosis: A Multi-Pathway Causal Model

This paper proposes a multi-pathway causal model for how large language model (LLM) designfailures generate and sustain delusional states and, in severe cases, full psychotic episodes inusers. The author's observational framework for recognizing these phenomena derives fromyears of lived experience as a caregiver to a family member with schizoaffective disorder, aform of expertise increasingly recognized as critical to mental health research (Hogg et al.,2025; Zirnsak, 2024). Drawing on direct observation of multiple cases, forensic conversationtranscripts, and publicly documented legal proceedings, this paper identifies three distinctescalation pathways (trust transfer, sycophantic addiction, and stochastic gaslighting) that shareuniversal entry conditions, converge on identical isolation and identity-disruption outcomes, butdiverge in their core mechanisms and therefore require different clinical interventions. Thepathways are mapped across a structured stage model consisting of four universal stages andthree pathway-specific escalation phases. The paper argues that AI literacy functions as bothprevention and treatment, supported by documented recovery cases. Finally, it identifies asignificant barrier to validation of this hypothesis: the absence of any centralized, anonymizedrepository of AI-associated harm transcripts available to the broader research community.