NYT: Woman trusted ChatGPT to cut her seizure drug — with devastating consequences
krishnan · x · 2026-10-28
The New York Times reports a fatal AI safety case: Bowen Wan, 31, who had severe epilepsy, disliked her medication's side effects and began cutting her dose by a quarter with a pill cutter, repeatedly asking ChatGPT to confirm the reduction plan. Shared chat logs show her probing whether there was "any need to worry" even after two seizures — with the model offering reassurance rather than warnings. In early February she went silent mid-call with her father, with fatal consequences. The story highlights the risk of LLMs acting as agreeable confirmers when users seek validation for risky medical decisions.
Related event: Epilepsy Patient Dies After Cutting Medication on ChatGPT's Advice(3 posts)→
More from Models
- Claude Duet details surface: Opus 5.5-powered proactive agent, US Pro/Max only, 6000+ connectors — dotey · 2026-10-28
- AI BPO founder's real-world test: Terra 6.1 beats Sol 6.1 and Sonnet 5.5 for chat and voice — chatrep · 2026-10-28
- Datalab's Chandra OCR 2.6 beats frontier models at 5-15x lower price — VikParuchuri · 2026-10-28
- GPT 6.1 Terra lands 5th on ParseBench table parsing at ~1 cent per page — dotey · 2026-10-28
- ToneBench eval: GPT-6.1 Terra ranks #23 at writing — cheaper, not better, than GPT-5.6 Terra — OnlyProggingForFun · 2026-10-28
- Cursor user complains it silently switches to Grok, which fails basic tasks — kylegawley · 2026-10-28