ChatGPT has a new Health section: is it safe for users?

OpenAI has launched a new Health section in ChatGPT, which is focused on working with users' health data. It is positioned as a separate secure environment for analysis, activity metrics, and medical history, rather than just another update to the chatbot interface. Habr reports this here.

The idea of the section is to replace the chaotic search for symptoms online with more balanced recommendations. Users can provide the model with access to medical data and receive generalized advice without diagnosis. For this, Health legally integrates with fitness and health services, including Apple Health and MyFitnessPal.

ChatGPT Health: what the new section offers

Health is conceived as a separate container within ChatGPT, where users' medical information is stored. In this space, different types of data can be aggregated and used for contextual advice.The main capabilities of the section include:

• connection of external health and fitness apps;

• analysis of exam results and activity metrics;

• receiving recommendations without direct medical intervention.

Safety of ChatGPT Health and access restrictions

Data protection has become a key issue for developers. According to OpenAI CEO Sam Altman, data from Health is not used to train models and does not enter regular chats. Context can only be requested as needed, without automatic information sharing.

To reduce the risk of incorrect advice, more than 260 doctors were involved in the creation of the section and a special benchmark, HealthBench, was applied. At the same time, the service clearly states that it does not replace a doctor's consultation. Currently, access to Health is only available through a waiting list and is not accessible to users from Europe.

It is worth noting that now lawyers use ChatGPT's history as evidence in court.