ChatGPT now reads health data, just not in Germany

ChatGPT now reads health data, just not in GermanyOpenAI launched its ChatGPT Health feature on July 23, 2026, for all users in the U.S. Anyone in the U.S. who is logged in and at least 18 can connect medical records and fitness apps to the chatbot, regardless of whether they have a free account or a paid subscription. The feature is not available in Germany, the rest of the European Economic Area, Switzerland, or the United Kingdom.
What’s behind ChatGPT Health
The feature consolidates health data in one place. Users in the U.S. can link their medical records from hospital systems such as Epic or Oracle Health, as well as Apple Health, MyFitnessPal, or Weight Watchers. ChatGPT then answers questions based on the user’s own data, such as interpreting lab results or preparing for a doctor’s appointment. According to OpenAI, health-related chats are stored in a separate, additionally encrypted area and are not used to train the models. Another new feature is that the health context can now be accessed in any chat, rather than only through a separate menu. OpenAI now counts around 300 million health-related questions per week; in January, that figure was 230 million.
Why Germany is left out
Even when the service first launched in January, OpenAI excluded the European Economic Area, Switzerland, and the United Kingdom. The reason lies in the stricter regulations governing health data. Access to actual medical records is only available in the U.S. anyway. For German users, this means that the health mode using their own data will not be available for the time being. However, ChatGPT can still be used here for general health questions.
A lawsuit casts a shadow over the launch
One day before the service went live, a pastor from Florida filed a lawsuit against OpenAI. According to a report in The New York Times, ChatGPT allegedly advised him not to see a doctor, with nearly fatal consequences. In its response, OpenAI cites the terms of service, which state that the service is not intended for diagnosis or treatment. The company emphasizes that it is working with doctors to provide safer responses, and that the smallest model of the latest generation, GPT-5.6-Luna, performs better than its predecessor in the company’s in-house HealthBench test. Nevertheless, several studies warn that AI chatbots are an unreliable source of medical advice.
What this means for you
Even without ChatGPT Health, many people have long been using the chatbot for health-related questions. This can help with interpreting test results, but it is no substitute for a medical evaluation, especially in cases of serious symptoms. Anyone who enters sensitive health data should be aware that it is processed on the provider’s servers. We’ve summarized what AI services secretly store about you and how you can stop that.





