FerIQ

FerIQ Privacy Policy

Last updated: 23 May 2026

1. Who we are

FerIQ is developed by Ferhat Yigen, based in Denmark. The data controller for the limited data processing described in this policy is the developer. You can reach the developer at fyigen@gmail.com.

2. What data FerIQ handles

Voice recordings and transcripts

When you record audio with FerIQ, the audio file is saved to a private folder inside the app's sandbox on your device. Transcription is performed locally using the Whisper large-v3-turbo machine-learning model running on Apple Neural Engine. The resulting transcript, AI summary, action items, chapters, translations, and speaker labels are stored locally in your device's SwiftData database.

Audio compression

To save disk space, FerIQ automatically re-encodes recordings from raw CAF format to AAC (~10× smaller) in the background after each recording finishes processing. This happens on your device using Apple's CoreAudio APIs. No audio is sent anywhere.

Microphone

FerIQ requests permission to use your device's microphone solely to record audio for transcription. The microphone is only active while you are explicitly recording. We do not stream microphone audio anywhere.

Custom vocabulary

If you enter custom vocabulary (names, technical terms, brand names) in Settings, this text is stored locally on your device. It is intended to bias future transcriptions; today the input is saved but not yet applied to the transcription engine.

3. What we never collect

• We do not collect, log, or aggregate your recordings.
• We do not have a developer account system. There is no FerIQ login.
• We do not include analytics SDKs, crash trackers, or third-party tracking libraries.
• We do not know which features you use, how often you launch the app, or how long your recordings are.
• We do not track your device's location, contacts, calendar, photos, or any other system data.

4. iCloud synchronisation

If you have iCloud enabled on your devices, your recordings and transcripts may be synchronised between your iPhone, iPad, and Mac through Apple's iCloud service. This uses your private CloudKit database, meaning the data syncs between your own devices via your personal iCloud account. We, the developer, have no access to it.

Apple's Privacy Policy governs how Apple handles your iCloud data. You can disable iCloud for FerIQ in Settings → [Your Name] → iCloud → Apps Using iCloud.

5. Apple Intelligence features

The following features use Apple Intelligence (on-device Foundation Models on iOS 26 / iPadOS 26 / macOS 26 on supported hardware):

• AI summary — generates a written summary of your transcript
• Action items extraction — pulls tasks, decisions, and follow-ups from your transcript
• Chapters / topic detection — divides long transcripts into navigable topic-based chapters
• Translation — translates your transcript to one of 14 target languages

These features run entirely on-device using Apple's Foundation Models framework. Your transcript is processed locally and never sent to Apple's or our servers. If Apple Intelligence is not available on your device, the relevant feature is gracefully disabled with an explanatory message.

6. Speaker labels (FluidAudio)

If you enable speaker labels in Settings, FerIQ uses FluidAudio's offline diarization models to identify which speaker said which sentence. The first time you use this feature, FerIQ downloads ~100 MB of CoreML models from Hugging Face's CDN to your device. The models run entirely on-device — your audio is never transmitted to FluidAudio's, Hugging Face's, or our servers.

7. Whisper model download

On first launch, FerIQ downloads the Whisper large-v3-turbo speech-recognition model (~600 MB) from Hugging Face's CDN. This is a one-time download. The model is saved on your device and is used entirely locally for all transcription thereafter. Your IP address is naturally visible to Hugging Face during the download but no FerIQ-related identifier or recording content is sent — it is a plain HTTPS asset request.

8. App Store purchases

If you purchase a Premium subscription or Lifetime unlock, Apple handles the transaction via StoreKit. The developer receives only the purchase status (active / cancelled / trial) via Apple's APIs — we do not see your name, email, payment method, or any personally identifying information. Subscription management is handled entirely by Apple at apps.apple.com/account/subscriptions.

9. Your rights under GDPR

Because all your FerIQ data lives on your device (and optionally in your own iCloud account), you have complete control:

• Right of access: Open any recording in FerIQ to see all stored data for it.
• Right to rectification: Edit transcripts, summaries, and speaker labels directly in the app.
• Right to erasure: Delete individual recordings or use "Delete all my recordings" in Settings. iCloud copies on other devices will sync the deletion.
• Right to data portability: Export transcripts as plain text, Word, or PDF; export full recordings as zipped bundles including audio.
• Right to object / restrict processing: Toggle off any AI-powered feature in Settings.

Because we as the developer do not hold or process any of your recordings or transcripts, there are no further GDPR steps for us to take on your behalf — all rights are exercised directly through the app's UI.

10. Children

FerIQ does not knowingly collect personal data from children. The app is rated for ages 4+ on the App Store but contains no advertising, no in-app communication, no social features, and no analytics. Parents and guardians retain full control over their devices and their children's recordings.

11. Changes to this policy

If we materially change how FerIQ handles data, we will update this page and the "Last updated" date at the top. Material changes include adding any new data flow that involves sending data off-device. We commit to remaining a fully on-device app — any future change would be communicated clearly.

Changelog

• 23 May 2026 — Added sections covering AI summary, action items, chapters, translation (Apple Intelligence features), speaker labels (FluidAudio), expanded language support (14 transcription languages + 99 via auto-detect, 14 translation targets), and audio compression. Updated brand identity.
• 22 May 2026 — Comprehensive rewrite with detailed sections on each data flow.
• 18 May 2026 — Initial public release.

12. Contact

For questions about this policy, data requests, or any other privacy concern, email fyigen@gmail.com. We aim to respond within one business day.