Skip to main content
Dark green background, "Weak Application Security Can Cost You Millions," 3 slanted images of fingers pointing to digital locks, and a "Learn the Basics" button
A €805M Consent Lesson for Every Mobile App TeamLawful Basis for Processing
5 min readFor Privacy Officers

A €805M Consent Lesson for Every Mobile App Team

Italy's supervisory authority, the Garante, fined Meta €805 million over a Facebook for Android feature that suggested photos and videos from users' phone galleries. The feature was enabled by default, allowing the app to access photos, videos, and related metadata without explicit consent. Meta plans to appeal, calling the fine "disproportionate," but the compliance lessons are clear.

This checklist is for DPOs and privacy leads responsible for mobile consent design. It covers permission architecture, default settings, and jurisdiction. Each item has a clear done/not-done state and a description of what "good" looks like.

Prerequisites

Before you start, ensure you have:

  • A current inventory of every mobile app your organization ships, per platform (Android and iOS behave differently).
  • A list of every OS-level permission each app requests, mapped to the features that use it.
  • Engineering and product contacts who can change defaults and permission flows, not only legal contacts who can review them.

Checklist

1. Confirm which legal rule governs device access.
Reading data stored on a user's phone falls under the ePrivacy Directive, implemented in Italy through Article 122 of the Privacy Code and in other member states through their own national laws. The Garante's position was that access to what a phone stores requires free and informed consent, regardless of data transmission or harm. Ensure your consent analysis starts with the national ePrivacy implementation, then layers GDPR on top for the processing that follows.
Done when: each app's permission requests are documented against the ePrivacy rule, not only a GDPR lawful basis.
Good looks like: a one-page legal map per market showing which rule governs device access and which governs downstream processing.

2. Split bundled permissions by purpose.
The Garante found that Facebook asked for gallery access through a single permission covering the whole app and several functions with different purposes. A general notice about enabling other features appeared first but didn't clarify that users were also enabling gallery suggestions.
Done when: no single OS permission prompt or consent screen covers more than one distinct purpose.
Good looks like: each feature that touches device storage has its own request, its own explanation, and its own record of the user's answer.

3. Audit every default.
List every feature that touches device data and record its default state. Anything that reads gallery, contacts, location, microphone, or similar data should default to off until the user acts. The Garante's order requires the feature to be off by default, including for those who had already granted permission.
Done when: your default-state register shows "off" for every feature reading device-stored content, with sign-off from product.
Good looks like: defaults are reviewed at design stage, before release, and a new feature can't ship without an entry in the register.

4. Test whether an existing permission counts as consent.
An OS-level grant obtained for one purpose doesn't automatically cover a later feature. Meta argues the feature works only after users give permission on their device. The Garante's order shows that argument didn't hold where the grant was broad and the new purpose was unclear.
Done when: you've reviewed every feature launched after the original permission was granted and confirmed it was covered by what the user was told.
Good looks like: feature launches trigger a consent review, and where the purpose is new, you ask again with a specific request.

5. Write the feature-level explanation.
The order requires a separate, specific request that explains what the feature is for, that it accesses the gallery, and how to turn it off. Use that three-part test on your own screens.
Done when: each consent screen states purpose, the data accessed, and how to withdraw, in plain language.
Good looks like: a user can answer "what am I agreeing to, and how do I undo it?" after reading the screen once.

6. Make withdrawal as easy as granting.
Withdrawal must be as easy as giving consent (GDPR Article 7(3)). Point to the in-app toggle, not only the phone's system settings.
Done when: withdrawal takes no more steps than granting, and you've tested this on a real device.
Good looks like: a switch inside the feature itself, with the change taking effect immediately.

7. Keep evidence you can show a supervisory authority.
The Garante ran its own technical analysis and noted that neither the user nor the authority can verify which information a broad permission actually uses. Your records need to close that gap.
Done when: you can produce, per feature, what data is read, when, and what the user was shown.
Good looks like: consent logs tied to app version and screen text, plus technical documentation of what each feature actually reads.

8. Plan for the update gap.
The Garante treated the breach as continuing for people who hadn't updated the app and for those who had already granted permission. It judged Meta's post-complaint changes late and limited to users who update.
Done when: your remediation plan covers users on old app versions and users with legacy permissions.
Good looks like: server-side switches or forced re-consent flows, so a fix reaches everyone and doesn't wait for voluntary updates.

9. Reassess your jurisdiction assumptions.
Meta argued that Ireland's authority should handle the case. The Garante rejected this because the ePrivacy rules don't provide a one-stop-shop mechanism. Your lead supervisory authority under GDPR doesn't shield device-access questions.
Done when: your risk register lists each market where you have users, with the local ePrivacy enforcement route noted.
Good looks like: mobile consent changes are reviewed against local rules in your largest user markets, not only your main establishment's.

10. Build a response timeline.
The order set deadlines: 30 days to report on work started, 60 days to implement the changes, and 90 days to report on measures taken. Missing the 60-day mark means the feature can't be switched on for anyone without valid consent.
Done when: you have a documented plan showing how quickly you could turn a feature off and re-consent your users.
Good looks like: a tested kill switch and a rehearsed re-consent flow.

Common Mistakes

  • Treating the OS permission dialog as the whole consent story.
  • Bundling a new feature under an old permission because "the user already said yes."
  • Shipping a fix only in a new app version and counting on users to update.
  • Assuming the one-stop-shop mechanism covers ePrivacy device-access rules.
  • Relying on a general notice about "other features" that doesn't name the specific feature.

Next Steps

Start with items 2 and 3. A permission-by-purpose map and a default-state register take little time to build and quickly show where you're exposed. Then run item 4 against your last year of feature launches.

Consider the scale the Garante weighed: roughly 20 million Android users in Italy, a feature available since 2023, a 2019 prior decision against the company, and limited cooperation during the investigation. Aggravating factors accumulate. Most of them are within your control well before a complaint arrives.

Promotional banner for the Penetration Report Template Kit

You Might Also Like