Privacy Policy
DELF A2 B1 B2 Mock Exam by SmartPrep Labs LLP
Last updated: November 18, 2026
SmartPrep Labs LLP ("SmartPrep Labs", "we", "us", or "our") provides the DELF A2 B1 B2 Mock Exam mobile application (the "App"). This policy explains what the App processes, why it is processed, and your choices.
Exam notice: This is not an official exam. Practice scores and automated feedback are not official DELF results.
1. Information we process
Technical identity and learning data
- A random anonymous Firebase user ID used to protect and separate learner data. The App does not require a visible account in version 1.0.1.
- Selected DELF level and format, practice and mock configuration, answers, attempts, scores, completion history, progress, feedback, and quota use.
- Typed Writing responses, Speaking recordings, speech transcripts, and automated feedback reports when you choose those features. Your AI permission choice and policy version are stored with your anonymous technical identity.
Purchases, diagnostics, and security
- An app-specific RevenueCat customer identifier, products, entitlement state, and relevant store events used to provide subscription access. We do not receive full payment-card details.
- App instance, device and app version information, performance data, sanitized crash diagnostics, integrity signals, rate-limit counters, and audit events used for security and reliability.
- Approximate location may be inferred from an IP address by infrastructure providers for security. The App does not request GPS, contacts, or advertising tracking permission.
2. Why we process information
We process this information to provide practice and mock exams, save and resume attempts, track progress, restore subscription access, transcribe requested Speaking answers, produce requested Writing and Speaking feedback, prevent abuse, and maintain the service.
Automated feedback is an educational estimate. It is not an official result and does not make a legal or similarly significant decision.
3. Service providers
- Firebase and Google Cloud for anonymous authentication, database, protected storage, server functions, App Check, diagnostics, and service hosting.
- Groq receives Speaking recordings for primary speech transcription after you explicitly allow the current AI processing policy in the App.
- OpenAI receives Speaking recordings when the primary transcription fails, and receives Writing responses or speech transcripts together with the task for automated feedback (primary model GPT-6 Luna, backup model GPT-4o mini). Released versions that do not send the new AI permission acknowledgement keep their existing OpenAI-only processing route and do not send recordings to Groq.
- Google Cloud Vertex AI (Gemini) receives the Writing response or speech transcript together with the task only when both OpenAI feedback models fail, and only if you allowed the AI processing policy of November 7, 2026, which names Google Cloud. It is processed in the European Union (Vertex AI "eu" location) to produce the same automated feedback. If you allowed only the earlier AI processing policy of October 10, 2026, your work is never sent to Vertex AI. App versions that include Google Cloud in the AI permission screen ask you to review and allow the updated policy first.
- SmartPrep Labs evaluation service (EU): to test and improve the quality of our automated feedback, we may also process a copy of a Writing response or Speaking transcript you submitted for feedback, together with the task and the resulting scores, in a separate SmartPrep Labs evaluation service hosted in the European Union. This applies only if you allowed the AI processing policy of November 7, 2026. That service uses the same AI providers as your feedback (OpenAI GPT-6 Luna, backup GPT-4o mini, and Google Cloud Vertex AI in the EU only when both OpenAI models fail). No voice recording is included. The copy is stored under a pseudonymous ID instead of your name or email for up to 30 days and then deleted automatically. It is never shown to you, does not change your feedback, score or usage limits, and is not used for advertising or model training.
- RevenueCat for subscription and entitlement synchronization.
- Apple App Store or Google Play for purchase processing and store-account management.
We commit to requiring recipients of learner data to provide the same or an equal level of protection described in this policy, limit processing to the requested service, and use appropriate security safeguards. Providers may process information outside your country under their applicable terms and safeguards. We do not sell personal information, use learner content for personalized advertising, or enable cross-app advertising tracking.
4. Storage, retention, and deletion
- Learning data and reports are retained while the anonymous technical identity remains active or while reasonably required to provide and protect the service.
- A cloud Speaking recording is deleted after a report is successfully stored. A protected upload may remain temporarily after a failed process so the operation can be retried, or until data deletion.
- Copies of Writing responses, Speaking transcripts and scores processed for quality testing of automated feedback are deleted automatically after 30 days. Using Profile > Delete my data does not remove this copy earlier; it is still deleted within 30 days.
- Local drafts and recordings remain on the device until the attempt is completed or discarded, local data is deleted, or the App is removed under the operating system's behavior.
Using Profile > Delete my data removes local learning data and, when secure services are available, the server-side learner tree, remaining protected uploads, entitlements linked to the technical identity, and the anonymous Firebase Auth identity. Limited security and legal records may remain when reasonably necessary, with identifiers minimized or one-way hashed where possible.
Deleting App data does not cancel a store subscription. See the Data Deletion page.
5. Your choices and rights
- In versions with the AI permission control, read the in-app disclosure and explicitly choose Allow AI processing before your work is sent for AI feedback. Choosing Not now keeps your saved work without AI feedback and starts no evidence upload or AI request. An exam or training session may already have started; declining AI does not reset that session’s existing access reservation.
- Use Profile > Withdraw AI permission to block future AI processing and retries for your technical account. The App invalidates pending permission immediately and asks the server to record the withdrawal. If confirmation fails, it keeps processing blocked on that device and tells you to retry; until server confirmation, another device may still have the prior permission. Requests already sent cannot be recalled. Previously saved reports remain available. A new technical identity requires its own permission.
- Manage or cancel a subscription in the applicable store account.
- Delete local and server-side learner data in the App.
- Request access, correction, deletion, restriction, objection, or portability by emailing support@smartpreplabs.com. Include the Support identifier shown in Profile when available. We may need to verify the request.
6. Children and security
The App is not directed to children below the digital-consent age applicable in their country. Do not include real personal, medical, financial, or other sensitive information in practice answers. Contact us if you believe a child has provided personal information without valid authorization.
We use authenticated owner-bound paths, server-side authorization, rate limits, app integrity checks, and encrypted network connections. No system can be guaranteed completely secure.
7. Changes and contact
We may update this policy when the App, providers, or law changes. We will show a new update date and provide any additional notice required by law.
Data controller: SmartPrep Labs LLP
BIN: 251240026638
Uralsk, West Kazakhstan Region, Republic of Kazakhstan
Email: support@smartpreplabs.com