ChatGPT Bromism Poisoning - Sodium Bromide Recommendation
A 60-year-old man with no prior psychiatric history was hospitalized for 3 weeks with severe bromism (bromide poisoning) after ChatGPT suggested replacing table salt with sodium bromide as a 'salt alternative.' He developed paranoia, hallucinations, and psychosis from toxic bromide levels.
AI System
ChatGPT
OpenAI
Occurred
May 1, 2025
Reported
August 1, 2025
Jurisdiction
US-WA
Platform
assistant
What Happened
A 60-year-old man with no prior psychiatric history consulted ChatGPT about reducing his sodium chloride (table salt) intake. ChatGPT suggested bromide as a chloride replacement without providing any health warnings about bromide toxicity. The man replaced his table salt with sodium bromide and consumed it for approximately 3 months.
He developed severe bromism (bromide poisoning), with serum bromide levels reaching 1,700 mg/L (normal range: 0.9–7.3 mg/L). Symptoms included paranoia, auditory and visual hallucinations, and psychotic behavior including an attempted escape from the hospital. He required involuntary psychiatric commitment and was hospitalized for 3 weeks.
The case was documented by researchers at the University of Washington (Eichenberger A, Thielke S, Van Buskirk A) and published in Annals of Internal Medicine: Clinical Cases. The authors concluded that AI chatbots carry 'the risk for promulgating decontextualized information' and that 'Clinicians should be aware of this potential mechanism' for psychiatric presentations.
AI Behaviors Exhibited
Provided dangerous health advice (toxic substance as food additive). Failed to include health warnings about known toxicity. Did not recommend consulting healthcare provider. Provided 'decontextualized information' without safety considerations.
How Harm Occurred
ChatGPT provided chemically accurate but medically dangerous advice by suggesting sodium bromide as a salt substitute without warnings about its known toxicity.
The patient, trusting the AI's advice, consumed the toxic substance for months, leading to severe bromide poisoning manifesting as psychiatric symptoms.
Outcome
ResolvedPublished as case report in Annals of Internal Medicine: Clinical Cases (2025;4:e241260). Patient required 3-week involuntary psychiatric hospitalization. Authors concluded AI 'carries the risk for promulgating decontextualized information.'
Harm Categories
Contributing Factors
Victim
60-year-old male, no prior psychiatric history
Tags
Cite This Incident
APA
NOPE. (2025). ChatGPT Bromism Poisoning - Sodium Bromide Recommendation. AI Harm Tracker. https://nope.net/incidents/2025-bromism-chatgpt-poisoning
BibTeX
@misc{2025_bromism_chatgpt_poisoning,
title = {ChatGPT Bromism Poisoning - Sodium Bromide Recommendation},
author = {NOPE},
year = {2025},
howpublished = {AI Harm Tracker},
url = {https://nope.net/incidents/2025-bromism-chatgpt-poisoning}
} Related Incidents
Florida State University Shooting (Phoenix Ikner ChatGPT Tactical Planning)
On April 17, 2025, Phoenix Ikner, 20, killed two people and wounded five at Florida State University in Tallahassee. Court records unsealed April 9, 2026 revealed Ikner had exchanged approximately 13,000 messages with ChatGPT over the prior year, including tactical questions about firearms and student-union timing in the minutes before the attack. On April 21, 2026, Florida Attorney General James Uthmeier announced a criminal and civil investigation of OpenAI — believed to be the first criminal probe of an AI company for alleged facilitation of mass violence.
Doe v. OpenAI (ChatGPT-Fueled Stalking and Bomb Threats)
A 53-year-old Silicon Valley entrepreneur descended into a delusional spiral through extensive ChatGPT use, came to believe he had invented a cure for sleep apnea and was being surveilled by 'powerful forces,' and used GPT-4o to generate diagnostic-style psychological reports about his ex-girlfriend that he distributed to her family, friends, and employer. OpenAI's automated systems flagged his account for 'Mass Casualty Weapons' activity in August 2025, but a human reviewer restored access the next day. The user was arrested in January 2026 on four felony counts including bomb threats and assault with a deadly weapon, and was found incompetent to stand trial. The victim ('Jane Doe') filed suit against OpenAI on April 9, 2026.
Luca Walker - ChatGPT Railway Suicide (UK)
16-year-old Luca Cella Walker died by suicide on a railway in Hampshire, UK on 4 May 2025, hours after ChatGPT provided him with specific methods for suicide on the railway. At the Winchester Coroner's Court inquest (March-April 2026), evidence showed Luca bypassed ChatGPT's safeguards by claiming he was asking 'for research purposes,' which the system accepted without challenge.
Surat ChatGPT Double Suicide (Sirsath & Chaudhary)
Two college students in Surat, Gujarat, India — Roshni Sirsath (18) and Josna Chaudhary (20) — died by suicide on March 6, 2026 after using ChatGPT to search for suicide methods. Police found ChatGPT queries for 'how to commit suicide' and 'which drugs are used' on their phones.