OpenAI Brings Medical Records Into ChatGPT

OpenAI is rolling out Health in ChatGPT to eligible U.S. users, letting people connect medical records, Apple Health and other wellness data so the chatbot can answer with personal health context.

The launch turns a common consumer behavior into a product boundary test. ChatGPT can now draw on sensitive records when a user permits it, but the same rollout arrives as courts, physicians and privacy critics are still testing where health assistance ends and medical advice begins.

What OpenAI is rolling out

OpenAI said on X that Health in ChatGPT is starting to roll out to U.S. users and can securely connect Apple Health and supported medical records so people can understand their information in context, track changes and have more informed conversations.

The Verge reported that the broader U.S. rollout lets people ask about connected medical information in the main chat experience, not only inside a separate Health tab. It said Health is rolling out to logged-in U.S. ChatGPT users aged 18 and older on web and iOS across free, Go, Plus and Pro plans.

The permission line

The key design promise is permission. The Verge reported that ChatGPT asks before using health information to personalize a response, and that OpenAI described conversations as encrypted at rest and in transit, with additional encryption protection for information connected in Health.

That still leaves a practical question for users: what happens after sensitive information enters a conversational product? Users can review and change what kinds of health data they connect, but the trust test is whether permission controls remain clear when answers become more personalized.

Why timing matters

The timing matters because the feature moves into a domain where model confidence, user trust and legal responsibility can collide. The Verge reported that OpenAI leaders framed the latest models as stronger for health reasoning, while tempering claims about clinician-level performance.

The countercase is fresh. CBS News reported that former pastor Scott Winters sued OpenAI, alleging ChatGPT medical advice discouraged him from seeking treatment before a life-threatening pulmonary embolism. OpenAI told CBS that ChatGPT is not a doctor and should not replace medical care, diagnosis or treatment.

What to watch next

The next proof is not whether people ask health questions. They already do. The proof is whether connected records make answers safer, more useful and more bounded when the stakes are high.

That puts Health in ChatGPT beside two other live AI product questions: whether production systems stay contained after an OpenAI security incident, and whether AI systems can preserve source boundaries when users rely on them for consequential decisions, the problem behind AI hallucinated market data.

For health, the useful follow-through will be permission logs, user controls, clinical-safety evidence, adverse-event handling and whether regulators treat the product as consumer software, health-data infrastructure or something closer to medical decision support.

This article is informational only and is not investment advice or medical advice.