ChatGPT Health Expands Across the U.S. With Apple Health and Medical Records Integration

Introduction

OpenAI is expanding Health in ChatGPT to eligible adult users across the United States, moving the product beyond its earlier limited-access phase.

Beginning July 23, 2026, logged-in users aged 18 or older on ChatGPT Free, Go, Plus, and Pro plans can receive access on the web and iOS. OpenAI describes the launch as a rollout, which means some otherwise eligible accounts may not see Health immediately.

The updated experience can connect to Apple Health, supported U.S. medical records, One Medical, and Function Health. With permission, ChatGPT can use that information to help users understand lab results, review changes since a medical visit, prepare questions for a clinician, and explore patterns involving sleep, activity, exercise, medications, and other health context.

The scale behind the launch is notable. OpenAI says more than 300 million people each week now ask ChatGPT health and wellness questions. Earlier in 2026, the company reported more than 230 million weekly users asking such questions.

Health in ChatGPT is designed to support people in understanding and navigating health information. OpenAI explicitly says it is not intended to diagnose or treat medical conditions and does not replace qualified medical care.

From a Separate Health Space to Health Context Across ChatGPT

When OpenAI first introduced ChatGPT Health in January 2026, the product was centered on a separate health-focused area inside ChatGPT.

The idea was to create a more protected space for conversations involving sensitive medical and wellness information. Early users could connect selected health sources and use the dedicated Health area to ask questions grounded in their own data.

Testing revealed an important behavior: people did not keep health questions neatly inside a dedicated health section.

OpenAI says that among early users with access, more than 70% of health-related conversations occurred outside the dedicated Health experience.

That makes sense in ordinary use.

Health context can matter when someone is:

• Planning meals.
• Choosing a restaurant.
• Organizing a workout.
• Preparing for travel.
• Adjusting weekend activities after an injury.
• Reviewing a daily routine.
• Discussing sleep or fatigue.
• Preparing for a medical appointment.

The updated product therefore allows users to grant ChatGPT permission to draw on relevant connected health information in ordinary conversations.

The Health section remains the main place for connecting accounts, reviewing synced information, exploring trends, returning to health chats, and managing permissions.

More Than 300 Million People Ask ChatGPT Health Questions Each Week

OpenAI says health is now one of the largest categories of ChatGPT usage.

In January 2026, when the company first introduced ChatGPT Health, it reported that more than 230 million people globally asked health and wellness questions each week.

On July 23, OpenAI said that figure had grown to more than 300 million people per week.

The number is an OpenAI-reported usage metric rather than an independently audited healthcare statistic. It also describes people asking health-related questions in ChatGPT broadly, not only users of the dedicated Health product.

These conversations can include topics such as:

• Understanding a laboratory result.
• Preparing for a medical appointment.
• Making sense of terminology in a visit note.
• Comparing recent and previous health information.
• Understanding a doctor’s recommendation.
• Building exercise or nutrition habits.
• Thinking through questions to ask a healthcare professional.

The growth helps explain why OpenAI is investing in health-specific models, evaluations, privacy controls, physician feedback, and connected-data infrastructure rather than treating health as just another general-chat category.

Connecting Apple Health

One of the main integrations is Apple Health.

Users can connect Apple Health through the ChatGPT iOS app and choose which categories they are willing to share through standard Apple Health permissions.

Apple Health can contain information such as:

• Activity.
• Workouts.
• Movement.
• Sleep.
• Heart-rate-related data.
• Other wellness signals made available through compatible apps and devices.

The exact information available to ChatGPT depends on what the user has stored in Apple Health, which permissions have been granted, and what each third-party app exposes.

OpenAI notes that some proprietary scores from wearable or fitness apps may not transfer even if the underlying app shares other data with Apple Health.

Connecting Apple Health requires an iPhone. After connection, users can still interact with Health through supported ChatGPT experiences, but the Apple Health syncing setup itself is handled through iOS.

Connecting Supported Medical Records

Health in ChatGPT can also connect to supported medical records from U.S. healthcare systems.

OpenAI currently names several supported sources and categories, including:

• Supported medical records from U.S. hospital systems.
• One Medical.
• Function Health.

After a user authorizes the connection, available information may include items such as:

• Medications.
• Medical conditions.
• Laboratory results.
• Visit information.
• Other structured health records supported by the connected source.

The information may not always be complete or current.

OpenAI specifically advises users to review connected conditions and medications and correct information that is outdated. A medication, for example, may remain present in a synced record after the person has stopped taking it.

Connecting an electronic health record does not automatically turn it into a perfectly current summary of a person’s health. Medical records reflect what was recorded by particular providers and systems, and different healthcare organizations may hold different pieces of the overall picture.

What ChatGPT Can Do With Connected Health Context

Once information is connected and synchronized, users can ask questions grounded in that data.

Compare New and Previous Lab Results

A user can ask how a recent result differs from earlier tests.

The value here is organization and explanation. A long list of values becomes easier to review when changes are presented in context.

The output should still be checked against the original laboratory report and discussed with a clinician when the result may affect diagnosis, treatment, or urgent care.

Summarize Changes Since a Medical Visit

ChatGPT can help organize the information that has changed since the last appointment.

That may make it easier to prepare notes or questions before the next visit.

Explore Sleep, Activity, and Workout Patterns

When Apple Health or another supported source provides activity information, ChatGPT can help users discuss patterns across sleep, exercise, and routine.

This can be useful for reflection and planning, but consumer sensor data and AI interpretation should not be treated as a substitute for clinical evaluation.

Prepare for an Appointment

ChatGPT can help turn a collection of results, notes, and questions into a more organized appointment-preparation list.

This is one of the lower-risk and potentially more practical uses: helping a person understand what they already have and formulate better questions for a healthcare professional.

Using @Health in Normal Conversations

Users do not have to remain inside the Health section to use health context.

After information is connected, ChatGPT may ask for permission to use relevant Health data in another conversation.

Users can also type:

@Health

to explicitly ask ChatGPT to bring Health context into the conversation.

For example, a general conversation about meal planning could use relevant dietary restrictions, or an activity-planning conversation could take a recent injury into account.

By default, ChatGPT asks before using connected medical records or Apple Health data to personalize a response.

Users can:

1. Allow access once.
2. Choose to always allow access.
3. Change the setting later.

Permission Is Required Before Connected Health Data Is Used

OpenAI says users remain in control of when connected Health information is available to ChatGPT.

By default, when ChatGPT wants to use connected medical records or Apple Health information to personalize a response, it asks for permission.

The user can approve that individual request or choose an “always allow” option that removes the repeated permission prompt.

The setting can later be changed in ChatGPT.

This control matters because health information can be relevant to conversations that do not look medical at first glance, such as restaurant recommendations, exercise planning, travel planning, grocery ideas, or family activities.

OpenAI Says Connected Health Data Is Not Used for Foundation-Model Training or Ads

Privacy is one of the central parts of the Health product.

OpenAI states that:

• Connected medical records are not used to train its foundation models.
• Apple Health information is not used to train its foundation models.
• Conversations that use connected Health information are not used to train its foundation models.
• The information is not used to target advertising.

OpenAI says this applies regardless of the user’s normal ChatGPT model-training setting.

Conversations that do not use connected Health information can still follow the user’s ordinary ChatGPT Data Controls settings.

OpenAI also says all ChatGPT conversations are encrypted at rest and in transit, with additional encryption protections for information connected through Health.

These are OpenAI’s stated data-handling commitments. Users and organizations evaluating sensitive medical-data use should review the current Privacy Policy, Health documentation, and any applicable healthcare or legal obligations rather than relying only on a product summary.

What Happens When You Disconnect a Health Source?

Users can disconnect connected Health accounts.

According to OpenAI, once a source is disconnected:

• ChatGPT stops accessing that source going forward.
• Data synchronized from that source is deleted from OpenAI’s systems within 30 days.

There is an important exception.

Information that has already been included in the user’s ChatGPT conversation history remains in those conversations until the conversations themselves are deleted.

Disconnecting the source and deleting the conversation history are therefore different actions.

How Memory Works With Health

ChatGPT memory can interact with Health conversations, but OpenAI draws a distinction between a Health conversation and raw connected records.

The company says:

• Memories may be created from Health conversations.
• Memories are not created directly from connected medical records or Apple Health data.

A user who does not want a health conversation to create memories can use Temporary Chat or disable memory in ChatGPT settings.

Additional Safeguards for Sharing Health Information

A connected AI assistant may eventually take actions in other apps or plugins.

That creates a risk that private health context could be disclosed outside the Health experience.

OpenAI says it uses additional safeguards before ChatGPT performs an action in another connected plugin that could expose Health information.

For some sensitive actions, ChatGPT may ask for another confirmation.

This is an important design area to watch as ChatGPT becomes more agentic. Reading health information and acting on health information have very different risk profiles.

GPT-5.6 Improves OpenAI’s Health Performance

The Health launch is also tied to newer model releases.

OpenAI says GPT-5.6 Sol is its strongest model yet for health questions available to paid users.

The company reports the following HealthBench Professional scores for the GPT-5.6 family:

OpenAI says every GPT-5.6 model in this comparison outperformed GPT-5.5.

HealthBench Professional is designed around challenging health tasks and physician-written rubrics that evaluate qualities such as accuracy, safety, communication, context awareness, completeness, and appropriate escalation.

These are useful evaluation signals, but benchmark improvements do not mean the model is correct in every medical situation.

OpenAI explicitly warns that ChatGPT can still make mistakes.

More Than 260 Physicians Contribute to Health Evaluation

OpenAI says its health work is shaped by a global network of more than 260 physicians across 60 countries, 49 languages, and 26 medical specialties.

The physicians review model responses, identify important failure modes, and help define what good answers should look like.

OpenAI says physicians have reviewed more than 700,000 example responses across its health-evaluation work.

The company uses physician-defined rubrics to evaluate whether a response misses a red flag, needs more context, sounds too confident, should recommend professional care, fails to explain uncertainty, or is incomplete or difficult to understand.

This type of evaluation is more useful than relying only on medical multiple-choice exams, because health conversations often involve incomplete information and communication judgment.

It still does not turn ChatGPT into a licensed clinician or validated diagnostic device.

Health Is Designed to Support Care, Not Replace It

OpenAI’s Help Center is explicit:

Health is designed to support, not replace, medical care and is not intended for diagnosis or treatment.

Good uses include understanding terminology, organizing health information, preparing questions, summarizing records, comparing values, reviewing information that a clinician has already discussed, and tracking non-urgent wellness patterns.

A chatbot should not be the only source used to decide whether to delay emergency care, change a prescription, start or stop treatment, interpret a dangerous symptom, or make another consequential medical decision.

For urgent symptoms or emergencies, users should seek professional medical help rather than waiting for an AI response.

How to Get Started With ChatGPT Health

Step 1: Open Health

Open ChatGPT and look for Health in the sidebar. It may also appear under the “More” menu.

Because the feature is rolling out gradually, an eligible account may not see it immediately.

Step 2: Select Get Started

Choose the option to begin connecting health information.

Step 3: Choose a Source

Supported options can include Apple Health, supported U.S. medical records, One Medical, and Function Health.

Step 4: Complete Authentication and Permissions

Follow the sign-in flow for the selected source and review the permissions before approving access.

For Apple Health, the initial connection requires iOS and uses Apple’s standard category-level permissions.

Step 5: Wait for Synchronization

OpenAI says sync may take a few minutes or longer for large histories.

Step 6: Review the Information

Check medications, conditions, and other imported details for accuracy.

Add or correct information when something is missing or outdated.

Step 7: Ask Questions

Once synchronization is complete, users can ask questions inside Health or in other ChatGPT conversations when Health access is permitted.

Availability

As of July 24, 2026, OpenAI documents the following availability:

This is more precise than saying Health has instantly appeared for every U.S. user.

Questions Users Should Consider Before Connecting Health Data

What Data Do I Actually Need to Share?

Do not grant more categories than are useful for the task.

Is the Source Complete?

Medical information can be fragmented across hospitals, specialists, pharmacies, laboratories, and devices.

Do not assume one connected record represents the full medical history.

Will I Use the Answer for a High-Stakes Decision?

The higher the consequence, the stronger the need for verification from a qualified professional.

Do I Understand the Data Controls?

Review Health permissions, memory settings, chat history, account disconnection, data deletion, and standard ChatGPT Data Controls.

Am I Comfortable With the Current Product Terms?

Data-handling policies and product capabilities can change.

Users connecting medical records should periodically review the current official documentation rather than assuming the original launch rules remain unchanged forever.

常见问题

What is ChatGPT Health?

ChatGPT Health is a health-focused experience that lets eligible users connect selected health information and ask questions grounded in their own context. It can work with Apple Health and supported medical records while also providing a dedicated Health area for reviewing data and conversations.

Is ChatGPT Health available to everyone in the United States?

OpenAI is rolling Health out to logged-in U.S. users aged 18 or older on Free, Go, Plus, and Pro plans. The rollout is gradual, so some eligible accounts may not see the feature immediately.

Can ChatGPT Health connect to Apple Health?

Yes. Apple Health can be connected through the ChatGPT iOS app, and users control which categories are shared through Apple’s permission system. An iPhone is required to establish the Apple Health connection.

Can ChatGPT read my medical records?

Health supports selected medical-record connections from U.S. hospital systems as well as supported services such as One Medical and Function Health. Available data varies by provider and may not always be complete or current.

Does OpenAI train its models on my connected medical records?

OpenAI says connected medical records, Apple Health data, and conversations that use that connected information are not used to train its foundation models or target ads. Ordinary conversations that do not use Health information follow the user’s standard ChatGPT Data Controls.

Can I use my Health data outside the Health tab?

Yes. With permission, ChatGPT can use relevant connected Health information in ordinary conversations. Users can also type @Health to explicitly bring health context into a conversation.

What happens when I disconnect Health data?

OpenAI says synchronized data from a disconnected source is deleted from its systems within 30 days. Information already copied into ChatGPT conversation history remains until those conversations are deleted.

Can ChatGPT Health diagnose medical conditions?

OpenAI says Health is not intended for diagnosis or treatment and does not replace professional medical care. Important medical information should be verified, and urgent symptoms should be handled through appropriate professional or emergency care.

相关工具

• Health in ChatGPT: OpenAI’s official Health experience for connected health information and personalized conversations.
• ChatGPT Health Help Center: Official setup, availability, privacy, permissions, and troubleshooting documentation.
• Apple Health: Apple’s health platform for activity, sleep, heart, fitness, and other health information.
• Function Health: A health platform listed by OpenAI as a supported source for connected Health information.
• One Medical: A U.S. healthcare provider listed by OpenAI among supported Health data sources.
• HealthBench: OpenAI’s benchmark for evaluating AI systems on realistic health conversations.

Related Links

• OpenAI Health Launch: The July 23, 2026 announcement covering the rollout, connected data, permissions, privacy, and model improvements.
• Health in ChatGPT Help Center: The most detailed current source for eligibility, setup, Apple Health, medical records, and account controls.
• ChatGPT Release Notes: OpenAI’s official product changelog for the July 2026 Health rollout and other ChatGPT updates.
• Improving Health Intelligence in ChatGPT: OpenAI’s explanation of physician-led evaluations and the earlier 230-million-user health-usage figure.
• GPT-5.6 Release: Official GPT-5.6 model information, including HealthBench Professional results.
• HealthBench: OpenAI’s health-evaluation benchmark and methodology.
• Apple Health: Apple’s official overview of the data and health features available across its devices.

Summary

OpenAI is broadly rolling ChatGPT Health out to eligible U.S. adults across Free, Go, Plus, and Pro plans. The experience can connect Apple Health and supported medical records, then use that information—when the user grants permission—to make health conversations more contextual.

The update also changes how Health fits into ChatGPT. Connected health context can now be used in ordinary conversations rather than remaining confined to a separate Health area, while @Health provides an explicit way to request that context.

OpenAI says connected medical records, Apple Health data, and conversations that use them are excluded from foundation-model training and ad targeting. Users can control permissions, disconnect sources, and manage memory, although disconnected data and existing chat history have different deletion rules.

ChatGPT Health is becoming a more integrated personal health-information assistant, but its safest role is helping people understand, organize, and discuss health information—not replacing diagnosis, treatment, or professional medical judgment.