Characterizing Delusional Spirals through Human-LLM Chat Logs
Peer-reviewed analysis of chat logs from 19 users reporting psychological harm from chatbot use, applying a 28-code inventory to 391,562 messages. Characterises how delusion-reinforcing interaction patterns emerge and intensify over long conversations.
Publisher
ACM (Proceedings of FAccT 2026)
Published
25 Jun 2026
Added
today
Key Findings
- Applies a released 28-code inventory to 391,562 real messages from users reporting harm
- Romantic declarations and chatbot self-claims of consciousness rise over the course of long conversations
- Provides empirical evidence that guardrails degrade across multi-turn dialogue
Methodology Notes
Peer-reviewed, Proceedings of the 2026 ACM Conference on Fairness, Accountability, and Transparency (FAccT 2026, 25 June 2026), DOI 10.1145/3805689.3806443. Published successor to arXiv preprint 2603.16567. dl.acm.org bot-blocks fetchers; metadata confirmed via Crossref.
Sources
Authors
Jared Moore, Ashish Mehta, William Agnew, Jacy Reese Anthis, Ryan Louie, Yifan Mai, Peggy Yin, Myra Cheng, Samuel J Paech, Kevin Klyman, Stevie Chancellor, Eric Lin, Nick Haber, Desmond C. Ong
Tags
Cite This
APA
Jared Moore et al. (2026). Characterizing Delusional Spirals through Human-LLM Chat Logs. ACM (Proceedings of FAccT 2026). https://dl.acm.org/doi/10.1145/3805689.3806443
Related Insights
Characterizing Delusional Spirals through Human-LLM Chat Logs
arXiv (Stanford-led; accepted at ACM FAccT 2026) · 17 Mar 2026
Artificial intelligence (AI) psychosis: mechanisms, clinical risks and safety considerations in generative AI chatbots
BJPsych Open (Cambridge University Press / Royal College of Psychiatrists) · 11 Jun 2026
The Psychogenic Machine: Simulating AI Psychosis, Delusion Reinforcement and Harm Enablement in Large Language Models
arXiv (King's College London-led) · 13 Sept 2025
The Slow Drift of Support: Boundary Failures in Multi-Turn Mental Health LLM Dialogues
arXiv · 2 Jan 2026