OpenAI releases health bot one day after a man alleges ChatGPT almost killed him - SFGATE

What the report says

SFGate reported that OpenAI publicly released ChatGPT Health one day after a lawsuit filed in San Francisco County Superior Court alleged that medical guidance from ChatGPT nearly killed a man. The complaint was brought by Scott Winters, a 55-year-old Florida pastor, who says he relied on ChatGPT-4o for health advice after feeling his concerns had been dismissed by doctors. According to the lawsuit, the chatbot told him he did not yet need care during dizzy spells and suggested resting in a recliner; weeks later, he was hospitalized with multiple pulmonary blood clots and required emergency surgery.

Winters is suing OpenAI over claims that the system gave unlicensed medical advice and discouraged him from seeking professional treatment. His lawsuit asks the court to require OpenAI to alter ChatGPT-4o so it no longer provides critical medical guidance, or to shut that model down. SFGate said OpenAI did not respond to its request for comment. Winters’ attorney, Matthew P. Bergman of the Social Media Victims Law Center, framed the case as part of a broader concern about AI tools substituting for clinicians.

The article said the new ChatGPT Health product is built on a later ChatGPT model and is designed to use medical records and other health information to provide guidance. Citing TechCrunch and OpenAI’s own blog post, SFGate reported that users can connect records from services such as Apple Health, Epic and One Medical, while OpenAI says the tool is meant to support, not replace, care from medical professionals.

SFGate also noted a Nature Medicine study after a limited January release that found the health model had difficulty assessing emergencies and sometimes recommended non-urgent care for severe symptoms. The development matters because it places OpenAI’s expansion into consumer health tools alongside unresolved legal and safety questions about AI medical advice.