Apple Health and HealthKit are designed around privacy. Apple doesn’t see your health data. Your records are encrypted on your device and in iCloud using keys only you control. No advertising. No selling your health data.
This architecture is genuinely privacy-protective. But it has a significant gap: third-party apps.
When you grant a third-party app HealthKit access, that app can read the health data you’ve authorized — and once that data leaves HealthKit, Apple’s privacy architecture no longer governs what happens to it. The app’s privacy policy takes over.
This post covers what HealthKit permission categories actually include, what can be inferred from seemingly innocuous data, and how to audit what you’ve already shared.
How HealthKit Permissions Work
HealthKit is Apple’s framework for health and fitness data. Your iPhone’s Health app is the central repository. Apple Watch, other wearables, and health apps write data into HealthKit. Third-party apps request permission to read from it.
Permissions are granular. An app can request access to specific data types: steps, heart rate, sleep analysis, menstrual cycle data, blood glucose, body mass, and dozens of others. You approve or deny each request category.
In practice, many users approve permissions they don’t scrutinize carefully, or approve broad permissions during app onboarding and forget about them.
The permissions persist. An app you installed three years ago and haven’t opened since may still have access to your HealthKit data. Unless you revoke it, the access continues.
What Each Permission Category Actually Covers
Apple’s permission labels are sometimes less specific than the data they unlock.
Steps and walking/running data. Sounds innocuous. But step count patterns encode significant behavioral information: when you leave and return home, your work schedule, whether you’re commuting, whether your activity level has changed over time (potentially indicating injury, illness, or life change).
Heart rate. Resting heart rate, active heart rate, and heart rate variability (HRV) are increasingly understood as indicators of cardiovascular health, stress levels, and recovery. Insurance companies have shown interest in heart rate data. Longitudinal trends can indicate developing health conditions before symptoms appear.
Sleep analysis. Sleep stage data (time in bed, time asleep, REM cycles, sleep interruptions) combined with timing information creates a behavioral profile. Sleep patterns are correlated with depression, anxiety, substance use, and other conditions. When you sleep and for how long is also sensitive from a basic security standpoint.
Menstrual cycle data. Post-Dobbs, the sensitivity of menstrual cycle data has received significant attention. This data type includes cycle length, fertile window predictions, and symptom logging. Many period tracking apps specifically request HealthKit access to sync data across platforms, meaning data logged in one app can be accessed by others with HealthKit permission.
Respiratory rate. Elevated resting respiratory rate is associated with illness, anxiety, and respiratory conditions. Continuous respiratory rate monitoring (from Apple Watch) creates a detailed baseline that can reveal deviations.
Blood oxygen (SpO2). Blood oxygen data from Apple Watch or pulse oximeters connected to HealthKit. Relevant to respiratory and cardiovascular health.
Blood glucose. For anyone with or at risk for diabetes, this is highly sensitive health information. HealthKit can receive data from continuous glucose monitors (CGMs) like Dexcom or Libre. Apps with HealthKit access to blood glucose can see this data.
Lab results and clinical records. If your health system is connected to Apple Health, clinical records including lab results, diagnoses, medications, and immunizations may be stored in HealthKit. Third-party apps that request “clinical records” access can read this information.
The Inference Problem
The sensitivity of health data isn’t limited to its face value. Inferences can be significant.
A fitness app that reads your step count, sleep analysis, and heart rate can infer:
- Whether you have a sedentary or active lifestyle
- Your work schedule and commute patterns
- When you experience high-stress periods
- How your health metrics change over time (trending better or worse)
- Whether lifestyle changes (reduced steps, disrupted sleep, elevated resting heart rate) coincide with identifiable life events
An app that reads your menstrual cycle data, body temperature, and sleep data has the inputs for pregnancy inference — a particularly sensitive category of inferred health information.
The pregnancy inference risk is concrete. Multiple women have reported that period tracking apps surfaced pregnancy before they disclosed it — to the app’s prediction UI, and potentially to the app’s backend. HealthKit doesn’t prevent apps from making inferences; it just restricts what raw data they can access.
What Happens When Data Leaves HealthKit
Apple’s privacy protections cover HealthKit itself. Once data is shared with a third-party app:
The app’s privacy policy governs. Read it. Specifically look for: whether health data is shared with third parties, whether health data is used for advertising, whether health data is sold, and how long health data is retained.
Most apps are not HIPAA-covered entities. HIPAA protects health data held by healthcare providers, insurers, and their business associates. A wellness app, fitness tracker, or period tracking app is typically not a covered entity. This means they can do things with your health data that a hospital legally cannot.
De-identification is easier to claim than to do. Many apps state that they share “de-identified” or “anonymized” health data with research partners or advertisers. Health data is notoriously difficult to truly anonymize. Research has repeatedly shown that combining step count, sleep, location, and heart rate data across multiple dimensions re-identifies individuals even without name or address.
Data can be shared in acquisition. If a health app company is acquired, your data typically transfers to the acquirer. Health data has been part of fitness app acquisitions multiple times. The acquiring company’s privacy policy then applies.
How to Audit Your HealthKit Permissions
On iPhone:
- Open the Health app
- Tap your profile picture (top right)
- Tap “Apps” under Privacy
- Review each app listed and what data types it has access to
This shows you every app that currently has HealthKit access and the specific data types authorized.
What to look for:
- Apps you no longer use. Revoke access from any app you haven’t opened recently.
- Apps with broader access than their function requires. A food logging app doesn’t need heart rate access. A sleep app doesn’t need clinical records access.
- Apps with access to sensitive categories (menstrual data, blood glucose, clinical records) that aren’t specifically healthcare tools you actively use.
Revoking access: Tap the app in the list, then toggle off the data types you want to revoke. You can remove all access from an app at once.
Apps Worth Auditing Specifically
Fitness and workout apps. Many request access to workout data, heart rate, steps, and active energy. This is expected. Scrutinize whether they also request sleep or other categories.
Period tracking apps. Given the sensitivity of menstrual data, understand where this data is stored, in which country, and what happens to it if you delete the app. Specifically: does the app have servers in the US, and is it subject to domestic law enforcement requests?
Insurance and employer wellness apps. Some health insurance programs and employer wellness platforms request HealthKit access to verify participation or calculate incentives. This is health data flowing directly to entities with financial incentives related to your health status. Read the specific terms carefully.
Social fitness apps. Apps that let you compete with friends on step counts or workouts often integrate HealthKit. The social and competitive features create data sharing beyond the app itself.
AI health apps. A growing category of apps uses HealthKit data to provide AI-driven health insights. The AI processing may occur on-device or in the cloud, and the health data may be used to train or fine-tune the AI model.
Privacy-Respecting Approaches
Minimize what you share. Only approve HealthKit access that an app genuinely needs for the feature you’re using. When an app requests permission to read multiple HealthKit data types, each one should have an obvious, specific use.
Use Apple Health locally. Apple Health itself is a privacy-respecting repository. You can log data in Health without sharing it with any third-party app. The health data in Health, synced to iCloud through Health’s encrypted format, doesn’t go to third parties unless you share it.
Revoke before deleting. When you delete an app, revoke its HealthKit access first. Deleting the app from your iPhone doesn’t automatically revoke HealthKit permissions in all versions of iOS.
Export for your own records. Apple Health lets you export all your health data as an XML file (Health app → profile → Export All Health Data). This gives you a personal archive that you control — useful if you want your health records without the continuous sharing implied by third-party app access.
The Broader Point
HealthKit is architecturally better than most alternatives for health data. Apple doesn’t sell your health data, doesn’t use it for advertising, and encrypts it end-to-end when synced to iCloud.
The vulnerability is the permission layer. Every app you grant HealthKit access to receives a copy of that data under their own terms. The protections Apple provides don’t travel with the data.
Treating HealthKit permissions with the same care you’d give a medical records release is the right frame. Would you authorize a company to receive copies of your health records with no specific use restriction? That’s effectively what broad HealthKit permissions involve.
Auditing what you’ve already granted takes five minutes. For data this sensitive, it’s worth it.