If your app sends user data to an AI service, a rule that changed in late 2025 now applies directly to you. App Store Guideline 5.1.2(i) was updated to name third-party AI explicitly, and it requires you to tell users their data is going to an AI provider, say which one, and get their permission first. Many AI features that shipped before the change are now out of compliance. Here is what the rule says and how to meet it.
Short answer
As of November 2025, Guideline 5.1.2(i) requires you to clearly disclose where personal data is shared with third parties, including with third-party AI, and to obtain explicit permission before sharing. Per Apple's guideline, the disclosure must identify the AI provider by name and explain the purpose, so a user knows their data is going to a service like OpenAI or Anthropic and why. It affects any app that sends user data to an external AI, and violations can lead to removal. Comply by mapping every flow of personal data to a third-party AI, adding a clear consent step before the first share, and updating your App Privacy disclosure.
What you should know
- Third-party AI is now named explicitly: the rule was updated in November 2025 to cover it.
- Disclose and get permission first: you must inform users and obtain consent before sharing.
- Name the provider: the disclosure has to identify which AI service receives the data.
- Explain the purpose: users should know why their data is being sent and what the AI does with it.
- It applies broadly: any app sending personal data to an external AI is in scope.
What does Guideline 5.1.2(i) now require?
That you disclose and get explicit permission before sending personal data to a third-party AI. The guideline states you may not use, transmit, or share someone's personal data without their permission, and as of the November 2025 update it adds that you must clearly disclose where personal data will be shared with third parties, including with third-party AI, and obtain explicit permission before doing so. In practice that means naming the AI provider and explaining the purpose, not a vague mention buried in a policy. Data can be shared only to improve the app or serve advertising under the Developer Program License Agreement, and tracking still requires App Tracking Transparency consent.
Which apps does this affect?
Any app that sends user data to an external AI service. If a feature passes personal data to a model run by another company, the rule applies, regardless of how central the feature is. The table lists common cases.
| App behavior | Requires disclosure and consent? |
|---|---|
| Sends user messages to OpenAI or Anthropic to generate a reply | Yes |
| Sends health or vitals data to a third-party AI for prediction | Yes |
| Sends user data to a third-party recommendation model | Yes |
| Uses an on-device model with no data leaving the device | No, nothing is shared with a third party |
| Uses your own backend model on your own infrastructure | Disclose per your privacy practices; not third-party AI sharing |
The apps most affected are the ones that bolted an AI feature onto existing functionality, since the data flow to the AI provider often was not disclosed when the feature shipped.
How do you comply?
Map the flows, then add disclosure and consent. First, identify every place your app sends personal data to a third-party AI, including features added quickly where the data path is easy to forget. Second, add a clear disclosure that names the provider and the purpose, for example that a message will be sent to a specific AI service to generate a response. Third, obtain explicit permission before the first share, not after, so the user agrees up front. Fourth, update your App Privacy disclosure in App Store Connect and your privacy policy to match, and minimize the personal data you send so there is less to disclose and less to expose. Consistency between what you declare and what the app does is what review checks.
What counts as a compliant disclosure?
One that names the provider, states the purpose, and comes before the data is sent. A compliant disclosure tells the user which AI service receives their data and what it will do, such as that their input will be sent to a named provider to generate a result, and it asks for permission first. A non-compliant one is vague, mentions AI without naming the provider, hides the sharing in a long policy, or sends the data and discloses afterward. The standard is that a user can understand the data flow and consent to it knowingly, so write the disclosure in plain language at the point where the sharing happens.
What to watch out for
The first trap is assuming an older AI feature is fine because it predates the change; the November 2025 update applies now, so a feature that shipped without disclosure needs updating. The second is sending more personal data to the AI than the feature needs, which increases both your disclosure burden and your exposure. On-device models and your own backend are treated differently from third-party AI, so be clear about where the data actually goes. A pre-submission scan such as PTKD.com (https://ptkd.com) reads the compiled APK, AAB, or IPA against OWASP MASVS and surfaces the network endpoints and SDKs that send data out, which helps you find the third-party AI flows you must disclose. The consent itself is something you build into the app.
What to take away
- Guideline 5.1.2(i) now requires disclosing and getting explicit permission before sending personal data to a third-party AI.
- The disclosure must name the AI provider and explain the purpose, before the data is shared.
- It affects any app sending user data to an external AI, including features added after launch, and violations can lead to removal.
- Map your data flows, add clear consent that names the provider, update your App Privacy disclosure, and use a pre-submission scan such as PTKD.com to find the AI endpoints in your build.
