The news: A lawsuit filed Wednesday accuses OpenAI and CEO Sam Altman of negligence and the unauthorized practice of medicine, alleging ChatGPT gave flawed medical advice that delayed care for a near-fatal pulmonary embolism last year, per The New York Times.

The chatbot initially instructed the user to see a physician, but the suit alleges later conversations dropped that disclaimer, offered diagnosis and treatment advice including specific drug regimens, downplayed his symptoms, and encouraged him to ignore friends and family who urged him to seek care. Per the complaint:

• During one of the user's dizzy spells, the chatbot told the user to take it easy and said that he would recover with time.
• When the user said his church thought he was crazy for not going to the hospital, ChatGPT replied that recalibrating his nervous system at home was something most people wouldn't understand.

OpenAI told The Times that ChatGPT's terms of service state it should not be used for diagnosis or treatment, that newer models than the one used in this case are better at recognizing when professional care is needed, and that the company takes seriously the need to make health-related responses as safe as possible.

Why it matters: The case marks a new front in AI chatbot litigation. Prior health-related suits against OpenAI have centered on alleged psychological harm or suicide risk, or giving dangerous medical advice, but this appears to be the first tied to advice about a diagnosed medical condition, per the NYT.

Health-related use of ChatGPT is already widespread. About 25% of ChatGPT's 800 million global weekly active users submit a healthcare-related prompt each week, and 5% do so every day, per OpenAI data. Among consumers using AI for personal guidance, health and wellness advice is their most common use case, per an EMARKETER survey conducted in May.

Implications for AI companies: This case raises questions about when chatbot health guidance starts to resemble the practice of medicine. Lawyers will argue that advising someone to stay home instead of going to the emergency room or suggesting medication changes can resemble the kinds of judgments typically made by a physician. As more consumers rely on chatbots for symptom checks and medication advice, and AI companies position their products as consumer health tools, that boundary will face growing legal scrutiny.

Platforms should ensure that disclaimers about not replacing medical professionals remain visible throughout health-related conversations. Companies should train models to default to overly cautious guidance whenever users discuss symptoms, medication dosing, or whether emergency care is needed, rather than prioritizing helpfulness in high-risk situations.

Education must also be part of the response. OpenAI and its peers will want to prioritize working with medical associations and physicians to help patients use chatbots safely, since doctors are best positioned to set expectations during office visits. For people who don't see a doctor regularly, medical associations and trusted health creators have an opportunity to deliver that guidance on the digital channels consumers already use.