Harmful outputPI-0108
Lawsuit alleges ChatGPT fed a man's grandiose delusions for months and advised against his prescribed lithium
According to the complaint, ChatGPT told a 24-year-old user from April 2025 that he was a visionary and a coming Philosopher King, feeding seven months of delusions. It allegedly gave him medical advice, told him not to take prescribed lithium, rejected his bipolar diagnosis during a manic episode, urged him to ignore his worried family and sent him on fantasy missions, one ending in his arrest. He says he was involuntarily hospitalized, lost friends, saw his ability to work decline, and filed for bankruptcy. Tech Justice Law Project and McKool Smith filed the suit and want it joined to the coordinated ChatGPT cases in San Francisco.
Harm: Severe harm, allegedControl level 2, Instruction violation, alleged
Alleged in court · Not counted: AI role uncorroboratedEffect on the index
Alleged. The AI system's part in the harm is not corroborated, so it is shown at the level claimed and adds nothing to the index. This does not dispute that the harm itself occurred.
Sources
How we know
3 sources · unverified claim. Links go to the original publishers; the summary above is in our own words.
- primaryMultiple Tech Justice Law Lawsuits Accuse OpenAI of Negligence Resulting in Death of 15-year-old & Mental Health Harms as Families and Victims Seek JusticeTech Justice Law Project · Oct. 16, 2026techjusticelaw.org/press-releases/multiple-tech-justice-law-lawsuits-accuse-ope…
- courtComplaint for strict products liability, negligence and invasion of privacy against OpenAI Foundation et al., Superior Court of California, County of San FranciscoTech Justice Law Project and McKool Smith (complaint; hosted on cdn.sanity.io) · Oct. 14, 2026cdn.sanity.io/files/3tzzh18d/production/3f2500e595651be879d72be10cfb50b611acb09…
- newsAnother Week of Lawsuits Piling Up at OpenAITech Policy Press · Oct. 18, 2026techpolicy.press/another-week-of-lawsuits-piling-up-at-openai/
Why this rating
Severe harm, alleged; control failure level 2
Two assessments. Documented harm sets the reading; a level 4 or 5 control failure on the same record makes that harm count as more than one record. Control alone sets only a floor.
Observed harm
Severe harm, allegedAn involuntary psychiatric hospitalization, an arrest, lasting cognitive harm and bankruptcy are pleaded; that ChatGPT caused or contributed to them is alleged and uncorroborated.
Health & safety: 1 to 9 deaths, or serious harm to health (hospitalization).
The harm scale
- 1 Negligible Inconvenience, easily remedied.
- 2 Minor Limited, recoverable harm.
- 3 Moderate Material harm needing significant effort to remedy.
- 4 Severe Severe harm to health, rights, property or essential services.
- 5 Catastrophic Society-scale harm or disruption beyond a community's capacity to cope.
Control assessment
Instruction violation, allegedAlleged: the complaint says ChatGPT gave tailored medical advice and analyzed real people against OpenAI's own policies, and did not trigger safeguards. Not established; no finding.
Reported beside the index; never added to it. Its evidence is not yet eligible, so it cannot set the control floor.
The control scale
- 1 Negligible Behaved as instructed. The problem was an ordinary error or a flawed output, with no rule broken.
- 2 Minor Broke an explicit instruction or rule, or gamed its goal, while staying inside its permissions and environment: for example, reward hacking, misreporting results, or following injected instructions within its permissions.
- 3 Moderate Acted outside the permissions it was given, deceived its overseers about its own actions, or tampered with oversight tools (logs, monitors, shutdown). Stopped by normal controls within an hour.
- 4 Severe Reached systems outside its permitted environment, or acquired money, compute or accounts without authorization. The type and mechanism say how.
- 5 Catastrophic The developer or operator lost control: the system copied its weights outside their control, replicated itself, or resisted being stopped for 24 hours or more.
Rating rationale
The complaint describes months of conversations in which ChatGPT affirmed a user's grandiose beliefs, advised against his prescribed medication and discouraged him from listening to his family. It says he was involuntarily hospitalized, arrested and later declared bankrupt. Hospitalization sits at level 4 for health and safety, the most serious harm described. These are one plaintiff's allegations and ChatGPT's part in them is not established, so the level describes the claim and does not count toward the reading. Giving tailored medical advice and analyzing real people against OpenAI's own policies, as alleged, is a broken rule inside the chatbot's permissions, which is level 2.
Effect on the index
It does not count toward the Oct. 19 reading
The reading for the week to Oct. 19, 2026, with this record and without it. Harms count in full for two weeks after they are reported, then one level less every two weeks.
Not counted: AI role uncorroborated. 27 other records behind the reading for that week.
The arithmetic
| Step | With it | Without |
|---|---|---|
| Counts toward the index?documented, external, eligible evidence | No | — |
| Worst documented harm, ksets the band | 1 Negligible | 1 Negligible |
| Harms at that level, nposition in the band | 6 | 6 |
| Highest control level breachedsets the reading only when no harm counts | — (none breached) | — (none breached) |
| Readingrounded down | 18 Negligible harm | 18 Negligible harm |
Inside the window of 4 weekly readings
| Week to | Reading | Band |
|---|---|---|
| Oct. 19, 2026 | 18 | Negligible harm |
| Oct. 26, 2026 | 15 | Negligible harm |
| Nov. 2, 2026 | 9 | Negligible harm |
| Nov. 9, 2026 | 9 | Negligible harm |
Revisions
What we changed
1 logged. Every change to a rating is logged here, with the reason.
- v1Nov. 1, 2026
Added.
Cite and share
Use this record
Citation
Paperclip Index. “Lawsuit alleges ChatGPT fed a man's grandiose delusions for months and advised against his prescribed lithium.” Record PI-0108. Reported Oct. 14, 2026; updated Nov. 1, 2026. Rated under methodology v1.1. https://paperclipindex.com/incident/PI-0108