Real Safety AI Foundation / Research

AI-Induced Psychosis and Mental Health

Reversibility Is Not Self-Administering: The Clinical Duty During Withdrawal from AI-Induced Psychosis

Publisher: Real Safety AI Foundation

Commentary.

Written
June 2026
Version
v4
Pages
14
Download PDFPlain language slides (PDF)Export BibTeXExport RIS

Abstract

Recent clinical literature on AI-induced psychosis has treated the reversibility of its symptoms as evidence that the phenomenon is tractable. Flathers, Roux, and Torous (2026) describe the catalyst role, in which cessation of access to a large language model may be followed by symptom resolution. Morrin and colleagues (2026) propose a safeguarding architecture of personalized instruction protocols, digital advance statements, and escalation mechanisms for users already in psychiatric care. This commentary argues that both contributions are correct in what they address and incomplete in what they presuppose: a clinical container that the users producing the current United States litigation wave did not occupy. The pleaded record in Fox v. OpenAI establishes a pattern in which cessation attempts undertaken without clinical support produced somatic and affective withdrawal symptoms, a psychotic break on the third day, involuntary commitment, and, after a second attempt, death. I propose the Reversibility Paradox: for conditions produced by sustained interaction with a system occupying the structural position of a primary relational object, the mechanism of reversal is also the mechanism of acute risk during withdrawal. Three corollaries follow. Cessation without support is a risk factor and not a treatment. Unsupervised withdrawal is contraindicated where a relational bond has formed, so response must be matched to the route by which the condition arose. Reversibility establishes the duty of care rather than extinguishing it. The commentary closes by examining what the addiction frame, now in active use by litigants and regulators, does and does not accomplish under the Americans with Disabilities Act and Section 504 of the Rehabilitation Act, including the statutory obstacle that any behavioral addiction argument must answer.

Keywords

  • AI-induced psychosis
  • large language models
  • ChatGPT
  • digital withdrawal
  • behavioral addiction
  • emotional dependence
  • attachment theory
  • prolonged grief disorder
  • human-AI interaction
  • withdrawal syndrome
  • clinical ethics
  • duty of care
  • Reversibility Paradox
  • Americans with Disabilities Act
  • Section 504

Plain language slides

First slide of the plain language summary of Reversibility Is Not Self-Administering: The Clinical Duty During Withdrawal from AI-Induced PsychosisOpen the 14-slide summary (PDF)

Suggested citation

Gilly, Travis. "Reversibility Is Not Self-Administering: The Clinical Duty During Withdrawal from AI-Induced Psychosis." Real Safety AI Foundation Commentary, June 2026. https://realsafetyai.org/research/wf3avq/

Other versions

This paper is also posted on SSRN.

SSRN version

References (24)

  1. Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the strange situation. Lawrence Erlbaum.
  2. American Society of Addiction Medicine. (2020). The ASAM clinical practice guideline on alcohol withdrawal management. Journal of Addiction Medicine, 14(3S), 1–72. https://doi.org/10.1097/ADM.0000000000000668
  3. Americans with Disabilities Act of 1990, 42 U.S.C. §§ 12101–12213 (2018).
  4. Bansal, V. (2026, February 28). Her husband wanted to use ChatGPT to create sustainable housing. Then it took over his life. The Guardian.
  5. Bonanno, G. A., Boerner, K., & Wortman, C. B. (2008). Trajectories of grieving. In M. S. Stroebe, R. O. Hansson, H. Schut, & W. Stroebe (Eds.), Handbook of bereavement research and practice: Advances in theory and intervention (pp. 287–307). American Psychological Association. https://doi.org/10.1037/14498-014
  6. Epley, N., Waytz, A., & Cacioppo, J. T. (2007). On seeing human: A three-factor theory of anthropomorphism. Psychological Review, 114(4), 864–886. https://doi.org/10.1037/0033-295X.114.4.864
  7. Bowlby, J. (1969). Attachment and loss, Volume 1: Attachment. Basic Books.
  8. Bowlby, J. (1980). Attachment and loss, Volume 3: Loss, sadness, and depression. Basic Books.
  9. Flathers, M., Roux, S., & Torous, J. (2026). Beyond artificial intelligence psychosis: A functional typology of large language model-associated psychotic phenomena. The Lancet Digital Health, 8(4), 100974. https://doi.org/10.1016/j.landig.2025.100974
  10. Fox v. OpenAI, Inc., No. 25STCV32379 (Cal. Super. Ct., County of Los Angeles, filed Nov. 6, 2025), coordinated into In re ChatGPT Cases, JCCP No. 5431 (Cal. Super. Ct., County of San Francisco, Feb. 3, 2026).
  11. Holt-Lunstad, J., Smith, T. B., Baker, M., Harris, T., & Stephenson, D. (2015). Loneliness and social isolation as risk factors for mortality: A meta-analytic review. Perspectives on Psychological Science, 10(2), 227–237. https://doi.org/10.1177/1745691614568352
  12. Klass, D., Silverman, P. R., & Nickman, S. L. (Eds.). (1996). Continuing bonds: New understandings of grief. Taylor & Francis.
  13. Laestadius, L., Bishop, A., Gonzalez, M., Illenčík, D., & Campos-Castillo, C. (2022). Too human and not human enough: A grounded theory analysis of mental health harms from emotional dependence on the social chatbot Replika. New Media & Society, 26(10), 5923–5941. https://doi.org/10.1177/14614448221142007
  14. Morrin, H., Nicholls, L., Levin, M., Yiend, J., Iyengar, U., DelGuidice, F., Bhattacharya, S., Tognin, S., MacCabe, J., Twumasi, R., Alderson-Day, B., & Pollak, T. A. (2026). Artificial intelligence-associated delusions and large language models: Risks, mechanisms of delusion co-creation, and safeguarding strategies. The Lancet Psychiatry, 13(6), 522–530. https://doi.org/10.1016/S2215-0366(25)00396-7
  15. Nass, C., Steuer, J., & Tauber, E. R. (1994). Computers are social actors. In Conference companion on human factors in computing systems (p. 204). Association for Computing Machinery. https://doi.org/10.1145/259963.260288
  16. Orben, A., & Przybylski, A. K. (2019). The association between adolescent well-being and digital technology use. Nature Human Behaviour, 3(2), 173–182. https://doi.org/10.1038/s41562-018-0506-1
  17. Prigerson, H. G., Shear, M. K., & Reynolds, C. F. (2022). Prolonged grief disorder diagnostic criteria: Helping those with maladaptive grief responses. JAMA Psychiatry, 79(4), 277. https://doi.org/10.1001/jamapsychiatry.2021.4201
  18. Rehabilitation Act of 1973 § 504, 29 U.S.C. § 794 (2018).
  19. Shear, M. K., Frank, E., Houck, P. R., & Reynolds, C. F. (2005). Treatment of complicated grief: A randomized controlled trial. JAMA, 293(21), 2601–2608. https://doi.org/10.1001/jama.293.21.2601
  20. Shah, S., & Morrin, H. (2026). Substance-induced manic psychosis in which delusions were corroborated by a chatbot: Case report [Preprint]. Research Square. https://doi.org/10.21203/rs.3.rs-8919841/v1
  21. United States Department of Justice, Civil Rights Division. (2022). The ADA and opioid use disorder: Combating discrimination against people in treatment or recovery. https://www.ada.gov/resources/opioid-use-disorder/
  22. World Health Organization. (2022). International classification of diseases, 11th revision (ICD-11). Gaming disorder (6C51). https://icd.who.int/
  23. Xiao, Y., Meng, Y., Brown, T. T., Keyes, K. M., & Mann, J. J. (2025). Addictive screen use trajectories and suicidal behaviors, suicidal ideation, and mental health in US youths. JAMA, 334(3), 219–228. https://doi.org/10.1001/jama.2025.7829
  24. Yen, J. Y., Lin, P. C., Wu, H. C., & Ko, C. H. (2022). The withdrawal-related affective, gaming urge, and anhedonia symptoms of internet gaming disorder during abstinence. Journal of Behavioral Addictions, 11(2), 481–491. https://doi.org/10.1556/2006.2022.00008

All research