OpenAI has launched Health in ChatGPT to U.S. users, bringing Apple Health data and electronic medical records directly into the AI assistant, and it works right from a Windows browser—no iPhone needed once the initial connection is set up. The feature, which began rolling out on July 25, 2026, promises to turn fragmented health information into a coherent, conversational resource while keeping sensitive data under strict privacy controls.

A New Health Option in ChatGPT’s Sidebar

The Health feature appears as a new entry in ChatGPT’s sidebar or More menu on the web and iOS. Once enabled, it allows users to connect select data sources and ask natural-language questions about their own health information. Instead of memorizing lab values or scrolling through disjointed patient portals, you can ask ChatGPT to compare recent test results, identify medication changes, or simply explain what a medical term means—all while referencing your personal records.

OpenAI stresses that Health in ChatGPT is not a diagnostic tool. It is designed to improve understanding and preparation, not to replace a doctor. The most practical uses are educational and organizational: translating jargon, building appointment question lists, summarizing past visits, spotting trends in sleep or activity data from wearables, and compiling timelines of medical events.

The rollout covers Free, Go, Plus, and Pro plans for logged-in U.S. users aged 18 or older. It is not yet available on Codex or Android, and there is no timetable for international expansion.

Connecting Apple Health, Medical Records, and More

Setting up Health in ChatGPT starts in the sidebar. Users tap “Get started” and follow the guided steps to link supported accounts. The current connections include:

  • Apple Health (iOS only for the linking step)
  • Supported electronic medical records from U.S. hospital systems
  • One Medical
  • Function Health
  • Other wellness, fitness, or nutrition apps where available

Apple Health is the central pipeline for data collected by an iPhone, Apple Watch, and compatible fitness or sleep tracking apps. ChatGPT can only use what Apple Health actually receives, so if an app does not write a metric to Apple Health, or if you’ve disabled data sharing within Apple Health, the AI won’t see it.

Medical records connections may provide lab results, visit summaries, clinical history, medications, and conditions. But records are not always complete or up to date—a discontinued medication may still appear, recent labs may be pending, and diagnoses can be provisional. OpenAI explicitly advises users to review connected information and verify it against original clinical sources.

Using Health Context Outside the Dedicated Sidebar

One of the most significant design choices is that health data can inform answers in other ChatGPT conversations, not just the Health section. You can type @Health in a standard chat to pull in relevant context, or ChatGPT may ask permission when it detects that your connected data could help personalize a response.

Permission prompts are enabled by default, but users can switch to an always-allow setting. OpenAI warns that granting blanket permission might be convenient, but it removes a valuable privacy checkpoint—especially if you share devices or use ChatGPT for unrelated tasks. For most people, leaving prompts on is the safer option.

What the New Feature Means for Windows Users

The Health feature is a browser-based experience, so Windows users can access it fully from their PC—provided they have an iPhone for the one-time Apple Health linking. The workflow looks like this:

  1. On your iPhone, open the ChatGPT app and connect Apple Health via the Health section.
  2. Confirm that your desired apps are sharing data with Apple Health.
  3. Wait for synchronization to complete and review what’s been imported.
  4. Switch to ChatGPT on your Windows browser; all connected data and health-themed conversations will be available.

This split-device approach lets you do the heavy reading, timeline analysis, or appointment preparation on a larger screen and keyboard, while your iPhone and Apple Watch handle the passive data collection. It also means that even if you’re a Windows-only household with a company-managed PC, you can still leverage your personal iPhone data in ChatGPT’s web client.

For users who rely solely on Windows and Android, the feature is not yet accessible. OpenAI has not announced an Android timetable, so the current audience is distinctly iPhone-plus-Windows users.

The AI Models That Power Health Conversations

Free users get GPT‑5.5 Instant for health interactions, which OpenAI has tuned to better recognize urgent situations, ask follow-ups, and explain uncertainty. Paid subscribers can use GPT‑5.6 Sol, which the company says performs better on its HealthBench Professional evaluation—a set of physician-developed scenarios testing accuracy, safety, communication, and appropriate escalation.

These benchmarks are encouraging, but they don’t mean the AI is clinically validated. Real-world health data is messy, often incomplete, and full of subtle context that a benchmark cannot fully replicate. A chat model may excel at explaining a cholesterol trend but still miss a crucial symptom or medication interaction that a clinician would catch immediately.

Privacy: The Strongest Promise and the Small Print

OpenAI has drawn a hard line on privacy for Health data: connected medical records and Apple Health information, along with any conversations that use that data, are not used to train foundation models or target advertising. This protection applies regardless of your general ChatGPT training preference.

Other privacy measures include:

  • All ChatGPT conversations are encrypted at rest and in transit; health data gets additional encryption layers.
  • Safeguards prevent other plugins from inadvertently exposing health information without confirmation.
  • Disconnecting a linked account deletes its synchronized data from OpenAI’s systems within 30 days.

But there’s a crucial caveat: any health information that already appears in your chat history remains there until you manually delete that conversation. So if you’ve had a lengthy, detailed discussion about a sensitive lab result, disconnecting the data source won’t scrub that chat. You must delete it yourself.

For extra control, use Temporary Chat for especially sensitive queries, and review your memory settings. OpenAI confirms that memories are not created directly from medical records or Apple Health data, but Health conversations can still contribute to memory if the feature is enabled.

Practical privacy checklist:
- Link only the accounts you truly need.
- Keep permission prompts enabled unless you have a compelling reason to disable them.
- Don’t introduce unnecessary health details into unrelated chats.
- Use Temporary Chat for discussions you don’t want retained.
- Periodically review and delete old conversations containing health data.
- Disconnect unused health accounts promptly.
- Verify that the data synced to Apple Health and through medical records is current and accurate.

Real-World Risks: Why Context Matters More Than Confidence

Health in ChatGPT reduces the friction of accessing and understanding medical data, but that same ease can mask serious risks. Three stand out:

  1. Automation bias – A fluently written explanation can feel authoritative, even when it’s based on an outdated medication list or incomplete lab history.
  2. Context collapse – A numeric lab value may be normal for one person and concerning for another, but the AI may lack the clinical nuance—pregnancy, recent procedures, or symptom evolution—that determines its real meaning.
  3. Urgency misjudgment – While models have improved at spotting red flags, a chatbot should never be the sole decision-maker in an emergency. OpenAI’s guidance is clear: if you suspect a serious problem, contact a medical professional or emergency services immediately.

There’s also a broader governance concern. Consolidating your most sensitive health information inside a consumer AI platform increases the stakes if a breach or unintended disclosure occurs. The privacy promises are strong, but no system is foolproof.

How We Got Here: From Generic Answers to Context-Rich Assistance

Health in ChatGPT didn’t emerge in a vacuum. For years, patients have struggled with fragmented data locked in multiple portals, PDFs, and wearable apps. Early generative AI could explain what a white blood cell count is, but it couldn’t relate that explanation to your actual results or your doctor’s notes.

OpenAI’s move signals a shift toward situated assistance—AI that works with your documents, your apps, and your ongoing context. The company built the feature with extensive physician input, developing the HealthBench evaluation to test nuanced, realistic scenarios rather than just flashy demos.

The rollout also reflects a growing understanding that health conversations don’t happen in a silo. Users were already asking health questions in standard ChatGPT chats; connecting data simply makes those interactions more informed.

What to Do Now: A Step-by-Step Guide

If you’re eligible and willing to try Health in ChatGPT, follow these steps:

  1. Check availability. The feature is rolling out to U.S. users aged 18+ on web and iOS. If you don’t see it yet, update your ChatGPT app and check back over the coming days.
  2. Link Apple Health on your iPhone. Open the ChatGPT app, go to Health, and tap “Get started.” Follow the prompts to connect Apple Health. Confirm which data categories you want to share.
  3. Connect medical records (optional). Look for supported providers in the same Health section and authenticate.
  4. Review imported data. Check that medications, conditions, and lab results are accurate and up to date. Remove stale items and add missing information like family history.
  5. Start with simple, organizational queries:
    - “Summarize my last three lab results and highlight any changes.”
    - “Create a list of questions for my upcoming checkup based on my medications.”
    - “Show me a timeline of my doctor visits over the past year.”
  6. On your Windows PC, simply log in to chatgpt.com. Your connected health data will be available, and you can continue conversations started on iOS.
  7. Adjust privacy settings. Leave permission prompts on, set up Temporary Chat for sensitive topics, and review your memory controls under Settings > Data Controls.

The Outlook: More Integration, More Caution

Health in ChatGPT is a careful first step, not a finished product. Expect future expansions to Android, additional U.S. healthcare systems, and possibly international markets once regulatory hurdles are cleared. Competitors like Google and Apple may respond with their own context-aware health AI features, but OpenAI’s cross-platform web access gives it a unique advantage—especially for the large Windows user base that also carries an iPhone.

The feature’s long-term value will depend on execution. Can it reliably synthesize messy real-world data without overstepping its bounds? Will users internalize the caveats, or will they start treating it like a virtual doctor? The answers will shape not just ChatGPT, but the entire next wave of AI assistants that promise to turn personal data into actionable insight.