Health in ChatGPT: What Changes When You Connect Your Data

Health in ChatGPT lets eligible U.S. users connect Apple Health and supported medical records so answers can use their own history. It can organize and explain information, but it does not turn ChatGPT into a doctor or make its output medically authoritative.

The feature is most useful for preparation: understanding a visit note, comparing results over time, finding questions for an appointment, or seeing how sleep and activity data relate to a routine.

What launched in July 2026

OpenAI began rolling Health out on July 23 to logged-in users aged 18 and older in the United States. At launch it is available on web and iOS across Free, Go, Plus, and Pro plans.

Users can connect:

• Apple Health;
• supported medical records from U.S. hospital systems;
• One Medical;
• Function Health.

Wearable, fitness, and nutrition apps may also contribute data through Apple Health when sharing is enabled. Available metrics vary, and proprietary scores may not transfer.

Health remains a dedicated area for managing accounts, records, trends, and past health conversations. With permission, ChatGPT can also use relevant connected information in conversations outside that area.

How the data flow works

By default, ChatGPT asks before using connected records or Apple Health information in a response. A user can approve one request or choose to always allow access. That setting can be changed later.

This distinction is important: connecting a source and allowing its data in every relevant conversation are separate decisions.

What it is useful for

The system may reduce the work of repeatedly finding and uploading the same information. It can also make a complex history easier to discuss. The output is still generated from records that may be incomplete, delayed, duplicated, or outdated.

OpenAI specifically warns that ChatGPT can make mistakes and does not replace qualified medical care.

Organizations building a separate knowledge assistant should treat this product as a reference, not an implementation blueprint. Our practical RAG guide explains how retrieval, source control, and permissions fit into a dedicated knowledge system.

Privacy controls that matter

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 normal model-training setting.

Other conversations that do not use Health data continue to follow the account’s standard Data Controls.

Important details:

• connected Health information receives additional encryption protections;
• users can disconnect a source at any time;
• OpenAI says synced source data is deleted from its systems within 30 days after disconnection;
• information already included in conversation history remains until those conversations are deleted;
• memories may be created from Health conversations, although OpenAI says memories are not created directly from connected medical records or Apple Health data;
• Temporary Chat or disabled memory can be used when a user does not want a health conversation to create memories.

Deleting a connection is therefore not identical to deleting every conversation that previously used its information.

A safer way to use it

Use Health as an explanation and preparation layer:

1. Review the synced conditions and medications before asking questions.
2. Correct anything incomplete or outdated.
3. Ask the system to separate recorded facts from possible interpretations.
4. Request the date and source behind every important statement.
5. Compare summaries with the original record.
6. Use the output to prepare questions, not to approve a diagnosis or treatment.
7. Delete sensitive conversations that should not remain in history.
8. Keep per-request permission enabled until the behavior is familiar.

A useful prompt is:

Use my connected Health information only for this response.
Create a dated timeline of the relevant recorded facts.
Separate facts from possible interpretations.
List missing or conflicting information.
Do not recommend changing medication or treatment.
Give me five questions to discuss with a qualified clinician.

When not to rely on it

Do not wait for an AI response during an emergency or use it to decide whether urgent care is necessary. Do not assume that missing information means a condition, medication, or risk is absent.

The feature is also a poor fit when:

• the connected record is known to be incomplete;
• several people share the same ChatGPT account;
• the user cannot review what was synced;
• a decision requires a physical examination or current clinical measurement;
• the information belongs to another person who has not consented.

The same least-privilege rule applies whenever an AI system can use external tools. The OpenAI–Hugging Face incident review shows why connected systems need isolation and explicit boundaries.

Bottom line

Health in ChatGPT is a significant shift from generic health questions toward answers grounded in a user’s own records and wearable data. Its best role is to organize, explain, and prepare.

The correct workflow remains: connect selectively, review the source data, allow access deliberately, verify important claims, and keep medical decisions with qualified professionals.