Your learning data is not fully local.

1. Who is responsible for your data

InfinLume is responsible for the personal information described in this policy. Questions and privacy requests can be sent to team@infinlume.com.

This policy is a notice about our practices; it is not a blanket request for consent. Where consent is required—for example, microphone or notification access—we ask through the app or your device.

2. Microphone recordings, transcripts, and AI

2.1 What happens to a recording

1. You choose to start a speaking activity and grant microphone permission.
2. The app creates a temporary audio file on your device.
3. The clip is sent over an encrypted connection through an InfinLume server function to a speech-to-text provider.
4. The provider returns text. InfinLume does not place the raw clip in our database or object storage.
5. We may store the returned transcript and speech-derived results with your account so the activity can be scored, shown in your history, troubleshot, and improved.

The temporary device file is subject to the app and operating system’s temporary-file lifecycle. Do not record sensitive personal information that is unnecessary for a learning exercise.

2.2 Speech-to-text providers

Important: “We do not store your voice” means InfinLume does not keep a raw-audio archive in its database or storage buckets. It does not mean a clip never exists temporarily on your device or at a transcription provider while the request is processed.

2.3 Transcripts and text-model providers

Transcripts are text, not raw audio. We store transcript text and speech-derived information such as expected wording, provider/model, confidence, acceptance result, score, duration, latency, and error details. Authorized personnel may review limited transcript and error data to investigate support issues, fraud, outages, and recurring recognition mistakes. We do not use this data to build voiceprints or to train an InfinLume AI model.

For some advanced speaking feedback, the transcript and exercise/evaluation context—not the microphone recording—may be sent to a text-model provider:

• OpenAI: OpenAI states that API inputs and outputs are not used to train its models by default unless the API customer opts in. Standard API abuse-monitoring logs may retain customer content for up to 30 days. OpenAI API data controls
• DeepSeek: The current scenario-evaluation implementation may use DeepSeek as a text-only provider and OpenAI as a fallback. DeepSeek’s public Open Platform terms do not give downstream users a no-training or fixed-retention promise; its general policy describes model training, an opt-out right, and processing in China. We therefore do not describe this text processing as zero-retention or no-training. Open Platform terms · Privacy policy

3. Information we collect

3.1 Information you provide or generate

• Account and profile: name, email address, avatar, sign-in provider, language goals, timezone, and profile preferences. Google or Apple provides only the account information you authorize; we do not receive those account passwords.
• Learning activity: lesson, quiz, flashcard, diagnostic, TCF practice, reading, listening, speaking, answers, scores, streaks, progress, history, and preferences.
• Speech-derived data: transcripts and the scoring, confidence, timing, provider, model, attempt, and reliability information described above.
• Communications: feedback, votes, support messages, waitlist submissions, and contact details you choose to provide.
• Subscriptions: product, entitlement, store, status, renewal/expiry, transaction identifiers, country/currency, and the account identifier needed to restore access. Apple or Google processes payment details; InfinLume does not receive your full card number.
• Notifications: device push token, permission state, timezone, quiet hours, reminder preferences, and delivery/troubleshooting records if you enable notifications.

3.2 Information collected automatically

• Technical and usage data: device, browser, operating system, app version/build, IP-derived approximate location, URLs or screens, timestamps, feature usage, performance, errors, clicks, referrers, and pseudonymous installation or analytics identifiers.
• Website analytics: PostHog, Google Analytics, and Vercel Analytics may measure visits and interaction. Selected diagnostic/results pages may use heatmaps or session recording for usability analysis. Claim-code text is masked where recording is used, and session recording is not intended to capture sensitive information.
• Cookies and local storage: continuity, preferences, anonymous TCF practice attempts and charts, diagnostic/claim identifiers, feedback identity, caching, fraud prevention, and analytics.

Our current mobile analytics disables automatic capture and mobile session replay. Mobile event properties are filtered for sensitive names such as authentication tokens, question text, translations, answer text, names, and email addresses.

4. Why we use information

• Provide, secure, troubleshoot, and maintain the Services.
• Sync and personalize learning, score activities, and show progress.
• Transcribe speaking practice and provide learning feedback.
• Validate subscriptions and restore paid access.
• Send notifications or communications you request.
• Measure performance, reliability, funnels, and which features are useful or confusing.
• Investigate abuse, enforce limits, reconcile provider billing, and comply with law.
• Improve content and deterministic scoring/recognition rules using individual support examples where necessary and aggregated or pseudonymized patterns where practical.

Depending on your location and the purpose, our legal basis may be performance of our agreement with you, legitimate interests such as security and product improvement, consent for optional permissions or communications, or compliance with law.

5. Field Notes, research, and learner data

InfinLume’s “Field Guide” and “Field Notes” are editorial names. They summarize published sources and practical language-learning methods. Reading them or using InfinLume does not enroll you as a research subject.

We do ordinary product analytics and quality assurance as described here. We do not publish identifiable learner data as editorial research. If we ever conduct a formal study involving users, we will provide a separate notice and obtain consent where required. A testimonial, quotation, or identifiable learner story is published only with permission.

6. When we disclose information

We do not sell personal information, rent it, or share it for cross-context behavioural advertising. We disclose only what is reasonably necessary to:

• Operate the service: Supabase (cloud database, authentication, storage, and server functions); Deepgram and Soniox (speech-to-text); OpenAI and DeepSeek (limited text processing); Expo (notifications); and PostHog, Google Analytics, and Vercel Analytics (analytics/hosting).
• Manage purchases: Apple App Store, Google Play, and RevenueCat.
• Support sign-in: Google and Apple when you choose their authentication service.
• Meet legal and safety obligations: comply with valid legal process; prevent fraud or harm; enforce agreements; or protect users, InfinLume, and the public.
• Complete a business transaction: a merger, financing, reorganization, or sale, subject to applicable safeguards and notice requirements.

7. Where information is stored and whether it syncs

• Supabase cloud: Our primary Supabase project is hosted in Canada (ca-central-1). It stores signed-in account, learning, subscription entitlement, transcript, and operational data.
• Your device: The app stores selected caches, settings, identifiers, and temporary files. Browser-only TCF practice progress and charts are stored in that browser’s local storage when you are not signed in.
• Other providers: Speech, AI, analytics, notification, authentication, and subscription providers may process information outside Canada. Current speech processing uses U.S. endpoints; DeepSeek text processing may occur in China.

Cloud-backed progress should be available after you sign in to the same InfinLume account on another supported device. Device-only settings and anonymous browser storage do not transfer automatically.

8. Retention and deletion

• Raw microphone audio: no InfinLume database or storage-bucket archive. Temporary device and provider processing is described in Section 2.
• Transcripts and learning data: generally kept while needed to operate your account, show history, score learning, and maintain reliability, then erased or de-identified according to a valid request and applicable law.
• Anonymous browser progress: remains until you clear infinlume.com site data or the browser removes it. It is not linked to the app’s account-erasure workflow.
• Operational, analytics, billing, and legal records: kept only as reasonably needed for security, fraud prevention, service limits, tax/billing reconciliation, disputes, legal obligations, or aggregate reporting.
• Backups and providers: deleted information may remain for a limited backup cycle or under a provider’s documented retention practices before being overwritten or de-identified.

Use Account Settings > Delete Account in the app to delete your InfinLume data. This removes account-linked learning history, progress, diagnostics, speaking transcripts and attempts, notification preferences, and local InfinLume app data, then signs you out.

We retain limited purchase, entitlement, and billing records where needed to validate or restore paid access. We may also retain minimal records needed for security, fraud prevention, accounting, legal compliance, dispute resolution, or to honour your communication choices.

9. Security

We use safeguards designed for the nature of the data, including encrypted network transport, access controls, row-level database authorization, restricted service credentials, and limiting raw-audio retention. No online service can promise absolute security. Protect your device and sign-in account, and contact us if you suspect unauthorized access.

10. Your rights and choices

Depending on where you live, including under Canadian, EEA/UK, or U.S. state privacy law, you may have rights to access, correct, delete, restrict or object to certain processing, receive a portable copy, withdraw consent, or appeal a denied request. These rights may have legal exceptions.

• Delete your InfinLume data in the app through Account Settings > Delete Account.
• Contact team@infinlume.com for access, correction, portability, objection, or appeal requests; we may verify your identity.
• Manage microphone and notification permissions in device settings.
• Manage an app-store subscription in Apple or Google settings; deleting InfinLume data does not cancel it.
• Block or clear cookies and local storage through browser settings. Some continuity and analytics features may stop working.

We do not discriminate against you for exercising an applicable privacy right. We do not sell personal information or use it for targeted advertising.

11. Children

The Services are not directed to children under 13, and we do not knowingly collect their personal information. If you believe a child under 13 provided personal information, contact us so we can investigate and delete it where required.

12. Changes to this policy

We may update this policy as the product, providers, or law changes. We will post the updated version, change the date above, and provide additional notice when required for a material change.