OpenAI launches ChatGPT Health with Apple Health and medical record connections

Why it matters

OpenAI's new health hub formalizes its stance on medical use of ChatGPT, aligning the product with emerging regulations and positioning the company to monetize a massive user base in a high‑risk, high‑value sector.

OpenAI launches ChatGPT Health with Apple Health and medical record connections

OpenAI is rolling out Health in ChatGPT to logged-in U.S. users age 18 and older, adding an opt-in system that can connect Apple Health and supported medical records to the assistant. The feature is launching on web and iOS across Free, Go, Plus, and Pro plans, but is not yet available in Codex.

The launch moves ChatGPT beyond answering isolated health questions. With permission, the assistant can use connected medications, lab results, medical visits, sleep, activity, and other health information to compare results over time, summarize changes since an appointment, or account for a dietary restriction or recent injury in an otherwise routine conversation.

OpenAI said more than 300 million people use ChatGPT for health-related questions each week. Those questions can range from interpreting lab results and preparing for appointments to understanding a doctor's explanation or developing a healthier routine.

Health context moves into regular conversations

OpenAI previously tested a dedicated health experience that required users to enter a separate area of ChatGPT. Among users with access, the company said more than 70% of health-related conversations occurred outside that space.

Health information can now be used across ChatGPT when relevant and permitted by the user. The dedicated Health area remains the control center for connecting accounts, reviewing recent data and trends, browsing synced records, managing permissions, and returning to earlier health conversations.

Users can connect Apple Health, supported medical records from U.S. hospital systems, One Medical, and Function Health. Data from wearables, fitness services, and nutrition apps can also reach ChatGPT when those services share information through Apple Health, although OpenAI cautions that available metrics vary and some proprietary scores may not transfer.

Connected records may also be incomplete or outdated. OpenAI advises users to review listed conditions and medications, remove information that is no longer relevant, add missing context such as family history, and verify important details against the original source.

OpenAI emphasizes permission and data controls

By default, ChatGPT asks before using connected medical records or Apple Health information in a response. Users can grant access for one request or choose to allow access without repeated prompts. They can also add @Health to a message to request that ChatGPT use their health context.

OpenAI said connected medical records, Apple Health information, and conversations that use those sources are not used to train its foundation models or target advertising, regardless of the user's broader model-training setting. The company also said all ChatGPT conversations are encrypted at rest and in transit, with additional encryption protections for information connected through Health.

Users can disconnect an account at any time. OpenAI said data synced from that source will be deleted from its systems within 30 days, but information already included in ChatGPT conversation history remains until the user deletes those conversations.

Memory is handled separately. ChatGPT may create memories from Health conversations, but OpenAI said memories are not created directly from connected medical records or Apple Health data. Users can turn memory off or use Temporary Chat to avoid creating memories.

The company also described additional checks for actions that could disclose health information through another connected service. ChatGPT may, for example, request confirmation before sharing a training plan based on Apple Health data.

New models and physician evaluations

OpenAI said it uses dedicated training and evaluations developed with hundreds of physicians to improve health responses. Those evaluations examine accuracy, safety, communication, completeness, awareness of context, and whether the model appropriately recommends professional care.

GPT-5.5 Instant, available to free users, is designed to improve recognition of situations that may require urgent care, request missing context, communicate uncertainty, and explain complex information. OpenAI describes GPT-5.6 Sol, available to paid users, as its strongest health model and says every GPT-5.6 model outperformed GPT-5.5 on HealthBench Professional.

The company said physicians also tested Health with connected data before launch. It nevertheless cautions that ChatGPT can make mistakes, does not replace qualified medical care, and should not be the sole basis for medical decisions.

What the launch means for health AI founders

For founders building health products, the consequential shift is not a new information page but the integration of longitudinal personal data into a general-purpose assistant used across everyday conversations. OpenAI is making medical records and wearable data part of ChatGPT's broader context layer rather than confining them to a standalone health product.

That approach creates a higher product and trust threshold for startups working on health summaries, appointment preparation, wellness coaching, and personal health navigation. OpenAI's rollout also highlights the operational details those companies must address: explicit permissions, stale or incomplete records, data deletion, memory controls, sensitive downstream actions, and clear reminders that an AI system does not replace a clinician.

The immediate rollout remains limited to adult U.S. users on web and iOS. Its usefulness will depend on the completeness of connected records, the reliability of the model's interpretations, and whether users are comfortable giving a conversational assistant continuing access to sensitive health context.