How to Keep Your Apple Health Data Private When Using AI Agents
Does ChatGPT Health send your heart rate to OpenAI? Yes. Here is how local-first MCP servers and on-device health AI keep your data private compared to cloud alternatives.
Quick answer: Your health data stays private when you use a local-first toolchain. An AI agent that never sends your data to the cloud, reads Apple Health read-only through a local MCP server, and answers questions on your own hardware. Cloud-based alternatives like ChatGPT Health or AI Health Export (Gemini) send your metrics to remote servers. If privacy matters, the architecture that never transmits anything is the only safe bet.
The cloud problem
Every time you ask a cloud AI about your health data, your metrics leave your phone. Not just the question. The actual numbers. Your resting heart rate. Your HRV readings. Your sleep stages. They travel over the network to a server you do not control, get processed by a model you cannot audit, and the company stores the result.
Most people assume that because the answer comes back in a chat interface, the data stayed local. It did not.
Here is how the major cloud health AI tools work under the hood.
ChatGPT Health launched in July 2026. You grant ChatGPT access to Apple Health via a data connector. Your metrics are uploaded to OpenAI servers. The company states it does not train on API data, but the data still transits their infrastructure.
AI Health Export (Gemini) sends your Apple Health data to Google Gemini cloud for processing. Every question you ask, every metric you query goes to a Google server. Google AI training policies have changed repeatedly in the last 18 months.
Health Auto Export runs a TCP-based MCP server on your Mac, but the iOS app must push data to a network destination. If that destination is a cloud folder or a webhook, your data sits on a third-party server.
There are now at least 10 Apple Health MCP servers on GitHub and npm. Most follow the same pattern: export your data to a shared folder or cloud destination, then expose it through an MCP server. Some require DuckDB. Some require Python. A few require Docker. Almost all of them send your data through a network layer at some point.
The local alternative
A genuinely private health AI for your Apple Health data does not send data anywhere. Full stop. Every step of the pipeline runs on hardware you control.
Here is what a local-first approach looks like.
The iOS app reads HealthKit read-only. Your metrics never leave the iPhone at this stage. A background export writes a compact JSON file to a destination you choose. That can be iCloud Drive (synced by the OS, no app developer in the path). A local folder on your Mac. A LAN share over your own network. A WebSocket endpoint you control, say over Tailscale. Or it can stay on your iPhone and only serve the on-device chat.
The MCP server runs as a single Node.js process on your computer. It makes zero network calls. It reads the JSON file from the local folder and exposes read-only tools to any MCP-aware agent. Claude, Cursor, opencode, ChatGPT Desktop with MCP support, any of them. The server has no telemetry, no analytics, no phone-home. It is 84 kilobytes on disk with zero dependencies.
When you ask a question, the agent calls a tool on your local MCP server. The server reads the data file from disk. The agent sees the result. Nothing left your network. No cloud account. No third-party server.
This is the architecture of MetricBridge. The iOS app handles the export. The open-source MCP server (health-export-mcp on npm and GitHub) handles the agent bridge. Together they form a pipeline that never sends your health data to anyone.
Two scenarios, one pipeline
You do not need to choose between the on-device chat and the MCP agent bridge. They work on the same data.
Scenario A: You just want to ask questions on your iPhone. Open the app. Type "how did my HRV trend this month?" The Guided engine computes the answer on device. You get a provenance card showing the exact samples, the date range, and the computation. No network call. No data leaves the phone.
Scenario B: You want Claude on your Mac to analyze your sleep patterns over the last year. The same iOS app writes the JSON file to iCloud Drive. The MCP server on your Mac reads it. Claude calls get_health_metrics and get_trends against your local data. The exact same privacy guarantee applies. Your data never left your iCloud account.
Both scenarios rest on the same principle: the data moves only through paths you control.
About the ad SDK. The App Store version of MetricBridge includes Meta ad attribution SDK. It sends three events (install, trial start, purchase) plus the advertising identifier after you grant App Tracking Transparency. No health data whatsoever reaches that SDK. This is enforced in code. The MCP server remains separate, open source, with zero network calls.
What the competitors do
AI Health Export (Gemini) is the most direct cloud competitor to MetricBridge. It lets you chat with your Apple Health data but uses Google Gemini in the cloud. Every answer requires a round trip to Google servers. There is no on-device inference and no MCP bridge. Your data leaves your phone.
Ask My Health is a newer iOS app with a built-in MCP server. It processes 100 free analyses on device then charges a monthly fee. The app claims no data collection but only covers workout and HRV metrics. It does not cover the full 190-metric HealthKit set. It does not show provenance cards. And without a Weekly Brief it does not surface trends proactively.
HealthMirror syncs your Apple Health data to iCloud Drive as JSONL files and provides its own MCP server. It is free but requires iCloud sync as a dependency. No AI chat. No provenance. No on-device question answering.
The on-device health AI apps (Risi, Metrya, Sifa, Thryve, Lifetrails, Ahimo, Suvi, Pharo, Priv AI) all process data locally. None of them have an MCP bridge. If you want your health data in Claude or Cursor, you need a separate export tool. None of them show full provenance on every answer.
MetricBridge is the only app that gives you both: on-device private chat with auditable answers, and a zero-dependency MCP pipeline to your own agents. Same data. Same privacy. Two surfaces.
Who this is for
The developer running Cursor who wants to query their HRV against their workout log. The quantified-self person who wants Claude to run weekly trend analysis. The privacy-conscious iPhone user who will not paste health data into a cloud chatbot. Anyone who has looked at ChatGPT Health and wondered "wait, does that send my heart rate to OpenAI?"
The answer is yes. ChatGPT Health sends your data to OpenAI. Every alternative that processes in the cloud does the same. The only way to be sure your Apple Health data stays private is to never send it anywhere.
That is exactly what a local-first, on-device architecture does. No cloud. No upload. No server in the path. For the step-by-step setup, see Connect Apple Health to Claude via MCP. To understand the export layer, see Export Apple Health data to JSON for AI.
Frequently asked questions
Does ChatGPT Health keep my Apple Health data private?
ChatGPT Health sends your Apple Health data to OpenAI servers for processing. Your metrics leave your device and transit their infrastructure. It is a cloud product, so the data necessarily travels over the network.
What is the most private way to use AI with my Apple Health data?
Use a local-first toolchain. An app that processes questions on device and an MCP server that makes zero network calls. MetricBridge plus its open-source MCP server is the only combination that meets this standard today.
Do any on-device health AI apps export to Claude or Cursor?
No on-device health AI competitor ships an MCP bridge. Risi, Metrya, Sifa, Thryve, Lifetrails, Ahimo, Suvi, Pharo, and Priv AI all process data locally but cannot feed that data to external AI agents. MetricBridge is the only app that does both.
Is MetricBridge really private?
The on-device chat processes everything on your iPhone. The MCP server makes zero network calls. The iOS app does include Meta ad attribution SDK that sends three anonymous events plus the advertising identifier after you allow tracking, but no health data ever reaches that SDK. Your health metrics, questions, and answers never leave your control.
Can I use MetricBridge without the ad tracking?
Yes. iOS shows an App Tracking Transparency prompt on first launch. If you tap Ask App Not to Track, the ad SDK sends nothing. The health features work identically either way.