ChatGPT Health Expands Across the US

The short version: Health in ChatGPT began rolling out across the United States on July 23. Adults can connect Apple Health and supported medical records, then let ChatGPT use relevant health context in ordinary conversations. OpenAI promises additional privacy controls, but the feature does not turn ChatGPT into a doctor—or place every interaction under the same law that governs hospitals.

What changed with the ChatGPT Health rollout?

OpenAI introduced a limited, separate ChatGPT Health experience in January. The new rollout makes connected health context available across ordinary ChatGPT conversations when the user permits it.

The company says more than 70% of health-related conversations among early users happened outside the dedicated Health area. Someone planning a meal might need to account for an allergy; someone arranging a weekend might want the assistant to consider a recent injury. Requiring those questions to begin in a separate space added friction.

Health remains the control center for connecting accounts, reviewing data and trends, managing permissions, and finding previous health conversations. Users can also add @Health to a prompt to explicitly request connected context.

Who can access Health in ChatGPT?

OpenAI says the rollout covers logged-in users who are:

• 18 or older;
• located in the United States;
• using ChatGPT on the web or iOS; and
• subscribed to Free, Go, Plus, or Pro.

The rollout is staged, so eligible users may not receive it simultaneously. OpenAI says the feature is not currently available in Codex.

Which health information can ChatGPT connect?

Users can choose to connect:

Source	Examples of available context
Apple Health	Activity, workouts, sleep, and other metrics shared by connected wearables or apps.
Supported U.S. hospital systems	Visits, medications, conditions, laboratory results, and clinical records made available through the connection.
One Medical	Supported account and care information.
Function Health	Supported laboratory and health information.

OpenAI warns that connected records may be incomplete or outdated. A medication, for example, may remain listed after the user stops taking it. Important details should be checked against the original record and corrected when necessary.

How does consent work when ChatGPT uses health data?

By default, ChatGPT asks before using connected medical records or Apple Health information to personalize a response. A user can allow access once or select an option that permits future access without repeated prompts. That choice can be changed under the Health plugin settings.

The distinction matters because Health context can now appear outside a clearly marked medical conversation. A restaurant recommendation, travel plan, or message composed through another connected tool could implicate sensitive information. OpenAI says additional checks apply before another plugin takes an action that could disclose Health data, and some actions may require confirmation.

The safest default for most users is per-request permission until they understand when Health context is being invoked.

Does OpenAI use ChatGPT Health data for training or advertising?

OpenAI says connected medical records, Apple Health information, and conversations that use that connected data are not used to train its foundation models or target advertising, regardless of the user’s general model-training setting.

That protection is specific. Conversations that do not use connected Health information continue to follow the user’s standard ChatGPT data-control choices.

The company also says:

• all ChatGPT conversations are encrypted in transit and at rest;
• connected Health information receives additional encryption protections;
• disconnecting a source deletes the synced data from OpenAI’s systems within 30 days; and
• information already included in conversation history remains until those conversations are deleted.

Health conversations can create ChatGPT memories, although OpenAI says memories are not created directly from connected records or Apple Health data. Users who do not want a conversation to create memory can use Temporary Chat or disable memory.

Is ChatGPT Health protected by HIPAA?

Not automatically. HIPAA regulates covered healthcare entities and their business associates; it is not a universal privacy law for every company handling health-related information. The Associated Press notes that sharing a medical chart with a consumer chatbot can involve different legal protections from giving it to a doctor or hospital.

That does not mean Health has no privacy obligations. Other federal and state laws, contracts, platform policies, and specific healthcare relationships may apply. It means users should not treat “medical information” and “HIPAA-protected information” as interchangeable categories.

OpenAI’s no-training, no-ads, encryption, permission, and deletion promises are material product commitments. They should still be evaluated alongside the precise terms, technical implementation, breach-notification rules, account security, and available remedies.

How medically reliable is ChatGPT Health?

OpenAI says it works with hundreds of physicians to create health scenarios and evaluation rubrics. It reports that GPT-5.6 models outperformed GPT-5.5 on its HealthBench Professional evaluation and describes GPT-5.6 Sol as its strongest model for complex health questions.

Those are OpenAI’s results. They do not eliminate the risk created by incomplete user descriptions, stale records, hallucinations, or a confident answer that mixes correct and incorrect advice.

Independent evidence remains more cautious. AP highlights research in which chatbot users did not make better decisions than people using search or their own judgment, even though models performed well when presented with complete written scenarios. Real people often omit crucial symptoms and struggle to separate good guidance from bad.

OpenAI itself says ChatGPT can make mistakes and does not replace qualified medical care.

Why is the timing controversial?

The broad rollout arrived one day after a lawsuit alleged that ChatGPT gave dangerous medical advice and discouraged a user from seeking care. The claims have not been adjudicated, and OpenAI disputes the characterization. The case nevertheless sharpens the central product question: what happens when an assistant designed to be helpful sounds more authoritative than its evidence warrants?

The answer cannot rely on a disclaimer alone. Health products need reliable emergency escalation, uncertainty communication, missing-context detection, auditable testing, clear consent, and interface choices that do not encourage medical dependence.

What should users check before connecting their health data?

• Use per-request permission until the access behavior is familiar.
• Review conditions and medication lists for stale or incorrect entries.
• Check which apps write data into Apple Health and which metrics they share.
• Decide whether health conversations may create memory.
• Delete relevant conversations as well as disconnecting the source when removing data.
• Protect the ChatGPT account with a unique password and multi-factor authentication.
• Verify important claims against original records and a qualified professional.
• Skip the chatbot and seek urgent medical help for emergency symptoms.

ChatGPT Health could make fragmented records easier to understand and help people prepare better questions. Its value will depend on whether users retain meaningful control—and whether the system reliably knows when explanation must give way to professional care.