Privacy & store compliance

Before you submit a mobile app that uses voice, you must declare the microphone and disclose that audio is sent to a third-party AI service for processing. This page is the copy-paste checklist for both app stores. It applies to the mobile app builds (App Store / Google Play) — the website widget ships no app and needs no store review.

The disclosure that matters

Voice is processed server-side on Gemini Live — no AI model runs on the device, but the user's audio leaves the device and is sent to a third-party generative-AI service. Both Apple and Google require you to disclose this. The SDK only captures/plays audio; the disclosure obligation is on your app's store listing.

What data leaves the device

Data	Where it goes	Notes
Microphone audio (voice sessions)	VoiceBot backend → Gemini Live	only while a voice session is active
Text turns	VoiceBot backend → Gemini Live	the typed message
App context (updateContext / sendCartEvent)	VoiceBot backend → the model	zero-PII by contract — see Context enrichment
Session token	VoiceBot backend	short-lived JWT, stored in Keychain / EncryptedSharedPreferences, never logged

The SDK does not ship its own analytics and does not collect end-user PII.

iOS

1. Microphone usage string (required)

NSMicrophoneUsageDescription — without it the app crashes on the first mic access, and review rejects it.

ios/.../Info.plist

<key>NSMicrophoneUsageDescription</key>
<string>Talk to the in-app assistant. Your voice is sent to our AI service to answer you.</string>

React Native

The Expo config plugin sets this for you; for a bare RN app add it to ios/<App>/Info.plist. You can override the Expo string:

app.json

{
  "expo": {
    "plugins": [
      ["@monoverse/voicebot-react-native/app.plugin", { "microphonePermission": "Talk to the in-app assistant." }]
    ]
  }
}

Flutter

Add it to ios/Runner/Info.plist (the plugin does not edit your Info.plist):

ios/Runner/Info.plist

<key>NSMicrophoneUsageDescription</key>
<string>Talk to the in-app assistant.</string>

2. Privacy manifest — PrivacyInfo.xcprivacy (required)

Apple requires a privacy manifest. Declare that audio data is collected and used for app functionality (and not for tracking). A minimal manifest:

PrivacyInfo.xcprivacy

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
  <key>NSPrivacyTracking</key>
  <false/>
  <key>NSPrivacyCollectedDataTypes</key>
  <array>
    <dict>
      <key>NSPrivacyCollectedDataType</key>
      <string>NSPrivacyCollectedDataTypeAudioData</string>
      <key>NSPrivacyCollectedDataTypeLinked</key>
      <false/>
      <key>NSPrivacyCollectedDataTypeTracking</key>
      <false/>
      <key>NSPrivacyCollectedDataTypePurposes</key>
      <array>
        <string>NSPrivacyCollectedDataTypePurposeAppFunctionality</string>
      </array>
    </dict>
  </array>
</dict>
</plist>

The React Native package bundles a privacy manifest for its own audio API usage. You still need a manifest for your app declaring your data collection.

3. App Review Guideline 5.1.2 — disclose audio → AI

Under Apple's data-use rules (Guideline 5.1.2), disclose in your App Privacy answers and your privacy policy that the user's audio is transmitted to a third-party AI service (Google Gemini, via the VoiceBot backend) to provide the assistant. Do not imply on-device-only processing.

Android

1. RECORD_AUDIO permission (required)

React Native

The Expo config plugin adds RECORD_AUDIO. For a bare RN app, add it to AndroidManifest.xml and request it at runtime before starting voice:

android/app/src/main/AndroidManifest.xml

<uses-permission android:name="android.permission.RECORD_AUDIO" />

Flutter

The package declares RECORD_AUDIO; the SDK requests it at runtime for you before capture (denial is non-fatal — text keeps working).

2. Data safety form

In the Play Console Data safety form, declare that the app collects audio and that it is sent to a third party (the VoiceBot backend / Gemini) for app functionality. Keep it consistent with your privacy policy.

3. Play Generative-AI policy

Because the assistant is powered by generative AI over the user's input, Google's Generative AI policy applies. At minimum:

• Disclose the AI-powered nature of the feature.
• Do not allow the feature to generate restricted/prohibited content; the VoiceBot backend applies prompt-safety, but your listing must still describe the feature accurately.
• Provide a way for users to report offensive output where applicable.

Scope note: foreground-only

v1 is foreground-only — there is no background audio, which deliberately avoids the iOS UIBackgroundModes: audio review path. Do not add background audio modes for this SDK.

Pre-submission checklist

• iOS NSMicrophoneUsageDescription set, and it mentions audio is sent to an AI service
• iOS PrivacyInfo.xcprivacy declares audio-data collection for app functionality, tracking = false
• iOS App Privacy answers disclose audio → third-party AI (Guideline 5.1.2)
• Android RECORD_AUDIO declared and requested at runtime
• Play Data safety form declares audio collection + sharing to a third party
• Play listing complies with the Generative AI policy (disclosure + content handling)
• No background audio modes added (foreground-only)
• AppContext / CartEvent payloads verified PII-free
• A reference build test-submitted to both stores before publishing the SDK to merchants