Messenger
Messaging · declared: 37 data types collected, 5 shared No tracking declared
Measured network activity crowdsourced
Observed in donated App Privacy Reports — measured on real iPhones by iOS itself, not declared by the developer. Every row below was reported by at least 3 separate accepted donations and published under a human verdict — rows marked inherited verdict arrived after a reviewer approved this app's evidence and inherited that decision. Contact counts and dates are aggregates across the donated reports.
| Contacted domain | Company | Classification | Donations | Contacts | Seen |
|---|---|---|---|---|---|
graph.facebook.com |
Facebook, Inc. 1 |
first-party | 4 | 23 | 2026-07-28 → 2026-07-29 |
chat-e2ee.facebook.com |
— | first-party | 4 | 15 | 2026-07-28 → 2026-07-29 |
web.facebook.com |
Facebook, Inc. 1 |
first-party | 4 | 8 | 2026-07-28 → 2026-07-29 |
How to read this
- A contact is a measurement, not an accusation. iOS records that the app (or code inside it, like an ad or analytics SDK) reached the domain. It does not show what data was sent.
- Counts reflect the donors' usage. A heavily-used app produces more contacts; compare domains within this app, not across apps.
- Separate donations, not proven separate people. Donations are anonymous by design, so we can count donations — we cannot prove three donations came from three donors. Human review weighs that before anything publishes.
- "Not in label" is the finding that matters. It means measured behavior with no matching disclosure in the app's own store privacy label — the declared-vs-observed gap this database exists to expose.
Domain owner/type attribution from Apple's App Privacy Report where the donor's export carried it; tracker classification via the AdGuard companiesdb (CC BY-SA 4.0). You can add to this evidence: donate your own App Privacy Report — parsed in your browser, shared only with your per-app consent, accepted anonymously (no account, no email, no IP address stored).
Think something here is wrong?
Evidence and research improve when people push back. Tell us what to re-check — a misattributed domain, an outdated classification, a claim about your own app — and a human will review it. Responses can take up to 30 days.