OpenAI launches ChatGPT Health for all in the US amid criticism

What happened?

OpenAI announced Thursday that ChatGPT Health is now available to all users in the United States. The feature, previously tested by a select group for several months, allows users to connect data from Apple Health, Function, and MyFitnessPal, as well as electronic medical records. Users can check recent trends, explore their history, and try suggested health-related questions.

According to the company, over 300 million people ask health-related questions on ChatGPT each week, and about one in three American adults has turned to a chatbot for health information in the past year, according to a KFF Health survey. This massive volume reflects a growing demand for accessible health information, but also raises questions about the accuracy and safety of the responses.

Why is this important?

The mass launch of ChatGPT Health represents a significant step toward integrating AI into everyday healthcare. OpenAI claims its latest models have improved their ability to accurately answer health queries, even outperforming doctors in some cases, according to a study published in Science that evaluated the o1 model. In that study, the o1 model demonstrated superior diagnostic accuracy compared to human doctors in simulated clinical scenarios, though the authors caution that these results do not directly translate to real-world practice.

Ashley Alexander, vice president of health products at OpenAI, called the expansion an urgent matter, noting that many people cannot access quality healthcare in a timely manner. “Everyone should have the right to this kind of support for their health,” she said. However, critics point out that urgency should not come at the expense of patient safety.

The historical context shows that AI in healthcare has advanced rapidly. In 2023, Google launched Med-PaLM 2, a model specialized in medical questions, but its deployment was limited to research settings. Microsoft, meanwhile, has integrated Azure Health Bot into hospital systems. ChatGPT Health, however, is the first generative AI product for health directly targeting mass consumers, making it a test case for regulation.

The shadow of a lawsuit

However, the launch comes amid controversy. On Wednesday, a lawsuit was filed against OpenAI and its CEO Sam Altman seeking an injunction to halt the operation of health-related products, including ChatGPT Health. Scott Winters, a 55-year-old pastor from Florida, alleges that an earlier version of ChatGPT gave him “personalized and inaccurate” medical advice that nearly caused his death from a pulmonary embolism in 2025. According to the lawsuit, Winters followed the chatbot's recommendations for treating chest pain symptoms, which delayed seeking urgent medical care.

The case underscores the risks of relying on chatbots for medical decisions. Although OpenAI has improved the accuracy of its models, the incident serves as a reminder that AI can still make serious mistakes, especially in contexts where the user's life is at stake. The lawsuit could set a legal precedent for the liability of AI companies in the healthcare domain, similar to medical malpractice cases against human professionals.

This is not the first such incident. In 2023, the National Eating Disorders Association shut down its chatbot Tessa after it gave harmful advice. And in 2024, a Stanford University study found that chatbots like ChatGPT and Bard (now Gemini) provided inaccurate information in 30% of medication queries.

Consequences and outlook

For users, ChatGPT Health offers a potentially useful tool for monitoring health trends and accessing information, but it does not replace professional medical judgment. Integration with Apple Health and other wearable devices facilitates personalized tracking, but raises questions about privacy and data security. OpenAI says data is handled with user consent and not used for model training, but the company has a history of changing its privacy policies, such as when it began using free user data for training in 2024.

For OpenAI, the launch is a double bet: on one hand, capturing a huge market (300 million weekly queries) and positioning itself as a leader in digital health; on the other, facing regulatory and legal scrutiny that could halt the initiative. The Winters lawsuit could set a precedent for AI company liability in healthcare, similar to how autonomous vehicle cases have shaped tort law.

Competitors like Google (with Med-PaLM) and Microsoft (with Azure Health Bot) are watching closely. Google has taken a more cautious approach, limiting Med-PaLM to controlled environments. Meanwhile, regulators may tighten rules for AI health products, requiring rigorous clinical testing before mass deployment. The FDA has already begun evaluating the need to regulate generative AI in healthcare, and this launch could accelerate that process.

The market impact is uncertain. If ChatGPT Health succeeds, it could spur other companies to launch similar products, increasing competition but also risks. If it fails due to legal or accuracy issues, it could slow AI adoption in healthcare for years.

What readers should know

• ChatGPT Health is available only in the US for now.
• It does not replace professional medical advice; OpenAI recommends consulting a doctor for health decisions.
• Personal health data is integrated with user consent, but privacy remains a concern, especially after OpenAI changed its privacy policies in 2024.
• The ongoing lawsuit could affect future availability of the service, and users should be aware that responses may be inaccurate.
• Model accuracy has improved but is not infallible; a Science study found o1 outperforms doctors in diagnoses, but only in controlled settings.

“Everyone should have the right to this kind of support for their health,” said Ashley Alexander, OpenAI's vice president of health, defending the urgency of the launch. However, the Scott Winters lawsuit reminds us that the right to information should not compromise patient safety.

In conclusion, ChatGPT Health represents a significant advance in democratizing health information, but also a risky experiment that could redefine the limits of AI in medicine. Users should use it with caution, regulators with attention, and the industry with responsibility.