OpenAI is taking ChatGPT deeper into personal healthcare by allowing users to connect medical records, lab results, medications and Apple Health data directly to the chatbot.
The new Health in ChatGPT experience is beginning its rollout to adults in the United States, giving ChatGPT permission-based access to a user’s health history so it can explain results, identify changes over time and help prepare questions for a medical appointment.
It’s a significant shift from general health advice towards personalised responses grounded in sensitive, real-world information. It also raises an obvious question: how much should people trust an AI assistant with both their medical data and their health decisions?
ChatGPT is moving beyond generic health answers
People have long used ChatGPT to ask about symptoms, medical terminology, test results, exercise plans and nutrition.
OpenAI says more than 300 million people now turn to ChatGPT with health-related questions each week. Until now, users often needed to manually copy information from medical portals, upload documents or repeatedly explain their history before receiving a useful answer.
Health in ChatGPT is designed to reduce that repetition. Once a user connects supported information, ChatGPT can consider details such as medications, previous laboratory results, recent medical visits, activity levels, sleep patterns and fitness data.
A user could ask it to compare a new test result with an earlier one, summarise changes since their last appointment or explore whether changes in sleep and activity appear alongside changes in their daily routine.
What health information can users connect?
The initial rollout supports Apple Health, selected medical records from US hospital systems, One Medical and Function Health.
Apple Health can also act as a bridge for compatible fitness, nutrition and wearable apps. When an app shares information with Apple Health, ChatGPT may be able to use the available data after the user grants permission.
Not every measurement will transfer. OpenAI warns that available information depends on the individual app, and some proprietary scores produced by wearable companies may not be passed through Apple Health.
The system is read-only. ChatGPT can view authorised information, but it can’t change a medical record, add a diagnosis or write information back to Apple Health.
Health information can now be used in regular chats
OpenAI originally introduced ChatGPT Health in January as a separate area within ChatGPT.
The dedicated section was intended to keep health conversations, uploaded files and connected services apart from ordinary chats. However, early testing revealed that people didn’t always think of their questions as strictly medical.
OpenAI says more than 70 per cent of health-related conversations among early users occurred outside the dedicated Health area. Someone planning dinner might want food allergies considered, while a user planning a family outing might need ChatGPT to account for an injury or mobility restriction.
The updated version can therefore draw on relevant Health information during other ChatGPT conversations, provided the user has granted permission.
Health remains available in the sidebar as the main place to connect accounts, browse records, review recent data and revisit previous health discussions. Users can also type @Health in a message when they specifically want ChatGPT to consider their connected information.
Users control when ChatGPT sees their data
By default, ChatGPT asks before using connected medical records or Apple Health information to personalise an answer.
Users can approve access for an individual request or choose to always allow it. That setting can later be changed within ChatGPT’s Health plugin controls.
Accounts can also be disconnected. OpenAI says information synced from a disconnected source will be deleted from its systems within 30 days.
There is an important exception. Health information already included in a previous ChatGPT conversation remains part of that conversation until the user deletes the chat itself.
Memories may also be created from things discussed during Health conversations, although OpenAI says memories aren’t generated directly from connected medical records or Apple Health data. Users can disable memory or use Temporary Chat when they don’t want a conversation remembered.
OpenAI promises stronger privacy protections
Medical information is among the most sensitive data a technology company can handle, making privacy central to the launch.
OpenAI says connected medical records, Apple Health information and conversations that use that data won’t be used to train its foundation models or target advertising. This applies regardless of the user’s normal model-training preference.
The company says all ChatGPT conversations are encrypted while being transmitted and stored, with additional encryption protections applied to information connected through Health.
OpenAI also says extra safeguards are applied before ChatGPT uses another connected service to share sensitive information. For example, it may seek confirmation before sending someone a training plan containing details drawn from their Apple Health data.
These protections are substantial on paper, but users will still need to understand what they’ve connected, what permissions they’ve granted and which conversations contain personal information.
ChatGPT Health isn’t an AI doctor
OpenAI repeatedly warns that ChatGPT Health is intended to support medical care, not replace it.
The company says the service isn’t intended to diagnose conditions or provide treatment. ChatGPT can still make mistakes, overlook important details or misunderstand incomplete information. Users are encouraged to verify important claims and discuss medical decisions with a qualified healthcare professional.
Connected records may also be incomplete or out of date. A medication could remain listed after someone has stopped taking it, while information held by one hospital or health app may not reflect care received elsewhere.
OpenAI advises users to review their connected conditions and medications, remove outdated information and check important details against the original source.
That limitation matters because a fluent, confident explanation can appear authoritative even when the data behind it is missing context.
ChatGPT may be useful for translating medical language into plain English or helping someone prepare questions, but it shouldn’t be treated as the final decision-maker.
OpenAI says its health models are improving
OpenAI says it has worked with hundreds of physicians to develop realistic health scenarios and evaluation standards covering accuracy, safety, communication, completeness and appropriate escalation to professional care.
The company says GPT-5.5 Instant has improved at recognising when urgent care may be necessary, asking for additional context and explaining uncertainty. GPT-5.6 Sol, available to paid users, is described by OpenAI as its strongest model for complex health questions.
OpenAI also reports that its GPT-5.6 models outperformed GPT-5.5 on HealthBench Professional, an evaluation designed around challenging medical conversations.
These are encouraging results, but they’re still evaluations rather than proof that an AI system will perform perfectly in every real-world case. OpenAI itself acknowledges that the models can make mistakes and require continued improvement.
A major step towards personalised AI
The launch shows where consumer AI assistants are heading.
Chatbots are becoming less dependent on information entered into a single prompt and more capable of drawing on persistent personal context, connected services and live data.
Health information could make an assistant far more useful. It could also make errors, privacy failures or poorly understood permissions far more consequential.
The success of ChatGPT Health may therefore depend on more than model intelligence. Users will need understandable controls, reliable integrations and clear boundaries between educational assistance and medical advice.
Who can access ChatGPT Health?
Health in ChatGPT is beginning its rollout to logged-in users aged 18 and over in the United States.
It is available on the web and iOS across ChatGPT Free, Go, Plus and Pro plans. Users with access can open Health from the ChatGPT sidebar or More menu and follow the instructions to connect supported accounts.
The current announcement doesn’t include an Australian release date. It also doesn’t confirm when broader international access or support for other national medical record systems will arrive.
Why this matters for Australia
Australians can already ask ChatGPT general health questions, but the new connected-record experience is currently limited to the United States. Local users won’t yet be able to connect Australian medical records through the feature announced by OpenAI. (OpenAI)
A future Australian rollout could be useful for people trying to make sense of information spread across doctors, hospitals, pathology providers, health portals and wearable devices. The value would depend heavily on which local systems were supported and whether users could clearly control what information ChatGPT accessed.
Privacy and accuracy would be just as important as convenience. Medical histories, medications and test results require a higher level of caution than an ordinary chatbot conversation, particularly when information may be incomplete or outdated.
For now, Australians should continue treating ChatGPT as a tool for explanations and appointment preparation, not as a substitute for a doctor, pharmacist or other qualified health professional.
Source
OpenAI
