ChatGPT can now reference connected health information across conversations when users allow it. That may include medications, lab results, recent visits, sleep, activity, workouts, conditions, and other records. OpenAI said the added context is intended to help users compare results over time, understand changes since an appointment, prepare questions for a doctor, and see how daily habits relate to their broader routine.
The company is positioning the feature as a way to organize and explain health information rather than replace professional care. ChatGPT can still produce incorrect answers, and OpenAI advises users to verify important details and discuss medical decisions with qualified providers.
Health remains available as a central management area in ChatGPT’s sidebar. Users can connect accounts, review synced records, browse recent data and trends, revisit past health conversations, and manage conditions or medications that may be incomplete or outdated.
Supported sources include Apple Health, medical records from participating U.S. hospital systems, One Medical, and Function Health. Data from wearables, fitness services, and nutrition apps may also become available when those services share information through Apple Health, although the specific information transferred depends on the app and metric.
OpenAI said users can ask health questions without connecting any personal information. Those who do link accounts can control when ChatGPT accesses the data. By default, the system requests permission before using connected records or Apple Health information in a response. Users can approve access once, allow it on an ongoing basis, or explicitly include the data by adding @Health to a message.
The company is also emphasizing privacy controls around the feature. Connected medical records, Apple Health information, and conversations that rely on that data are not used to train OpenAI’s foundation models or target advertising, regardless of the user’s broader model-training settings.
Users can disconnect connected health accounts at any time. OpenAI said data synced from a disconnected source is removed from its systems within 30 days, although information already included in existing conversations remains until those chats are deleted.
Health-related memories can still be created from conversations when memory is enabled, but OpenAI said memories are not generated directly from connected medical records or Apple Health data. Users can disable memory or use Temporary Chat when they do not want a conversation to contribute to future personalization.
OpenAI said it has also added safeguards for situations in which health information could be shared through another connected service. In some cases, ChatGPT may ask for confirmation before taking an action that could disclose sensitive data.
The rollout is tied to OpenAI’s broader work on health-focused model performance. GPT-5.5 Instant is available to free users, while GPT-5.6 Sol is positioned as the company’s strongest model for more complex health questions. OpenAI said the newer model performs better when reasoning across detailed records, explaining uncertainty, and helping users prepare for follow-up care.
The company said hundreds of physicians helped create evaluation scenarios and review the Health experience before release. Those assessments focused on accuracy, safety, communication, completeness, context awareness, and whether the model appropriately recognized when professional care might be needed.
OpenAI says more than 300 million people use ChatGPT for health-related questions each week. ChatGPT Health is designed to make those conversations more personalized by bringing together information that would otherwise remain spread across patient portals, apps, wearables, and medical records.
This analysis is based on reporting from OpenAI.
Image courtesy of OpenAI.
This article was generated with AI assistance and reviewed for accuracy and quality.
