DE EN

Privacy Policy.

Version 1.0 · Effective date: 24.04.2026.

00

Dikta is a privacy-first, on-device dictation and transcription app. This Privacy Policy explains what personal data Dikta processes, why, on what legal basis, and what your rights are. It is drafted to comply with the General Data Protection Regulation (Regulation (EU) 2016/679 — "GDPR") and the German Federal Data Protection Act (Bundesdatenschutzgesetz — "BDSG"), and applies to everyone who downloads or uses Dikta, anywhere in the world.

The short version: we were careful to build a product that does its work on your device. Audio and transcripts never reach our servers. The only data that routinely leaves your device is (a) anonymised crash reports you can switch off at any time, (b) subscription-lookup traffic when you use the paid tier, and (c) — if you are a paying subscriber and have signed into iCloud — your own transcripts and audio, mirrored into your own private iCloud account by Apple. The detailed version follows.

01
Controller

The data controller within the meaning of Art. 4 (7) GDPR is:

Name
Paul Lukas Roder
Form
Sole proprietor (Einzelunternehmer)
Address
Montessoristraße 21 40670 Meerbusch Germany
E-Mail
[email protected]
VAT
Application pending

We have not appointed a dedicated data-protection officer, because we are not required to do so under Art. 37 GDPR or § 38 BDSG. All privacy enquiries are handled by the controller directly at the e-mail above.

02
Scope

This Privacy Policy covers personal data processed through:

  1. the Dikta mobile, tablet, desktop, and spatial-computing app ("the App") on iOS, iPadOS, macOS, and visionOS;
  2. the Dikta home-screen, lock-screen, and StandBy widgets bundled with the App;
  3. in-app purchases and subscriptions handled through Apple's App Store;
  4. iCloud synchronisation of your transcripts, audio, and settings, available only to paying subscribers who have enabled iCloud;
  5. crash and error reports we receive when something goes wrong and you have not opted out;
  6. on-device summarisation performed through Apple's Foundation Models framework (Apple Intelligence); and
  7. the marketing website getdikta.com, insofar as it is operated by us.

It does not cover any third-party software or service you integrate the App with outside of the components described below.

03
Privacy by Design
  • On-device first. Dikta records audio, transcribes it with WhisperKit, and optionally diarises speakers with Pyannote — all entirely on your device. The audio and the resulting transcript never leave the device through any channel controlled by us.
  • No advertising, no tracking. Dikta contains no advertising SDKs, no ad identifiers (IDFA), no attribution SDKs, no cross-app tracking, no pixels, and no analytics that profile users.
  • No accounts. You do not create a Dikta account. There is no Dikta login and no Dikta-side user directory.
  • iCloud sync is opt-in and subscription-only. If you do not subscribe, your transcripts never leave your device. If you subscribe but are not signed in to iCloud, they still never leave your device. If you subscribe and are signed in to iCloud, Apple mirrors them into your own private iCloud database under your Apple ID — a data space we cannot read.
  • Speaker diarisation does not store voice biometrics. The diarisation engine produces short-lived numerical embeddings in memory to group segments by speaker; only the resulting speaker label (e.g. "Speaker 1") and the colour chip are persisted. No voice-ID, voiceprint, or biometric template is saved anywhere.
04
Data on Device Stays local

The following data is created and stored on your device. Unless you enable iCloud sync (see Section 5.3) or a crash occurs (see Section 5.2), it never leaves the device.

Audio recordings
Contents
Raw WAV — mono, 16 kHz, PCM16
Storage
Application Support directory (free) / SwiftData external-storage attribute (paid)
Basis
Art. 6 (1) (b) GDPR — contract performance
Transcripts and segments
Contents
Text output, word timings, confidence scores
Storage
Local SwiftData database (Dikta.local.store / Dikta.cloud.store)
Basis
Art. 6 (1) (b) GDPR
Speaker labels
Contents
"Speaker 1", "Speaker 2", … and colour codes. No embeddings, no voiceprints.
Storage
Same SwiftData database, linked to transcript
Basis
Art. 6 (1) (b) GDPR — only when diarisation is enabled
User metadata
Contents
Title, tags, favourite flag, flag timestamps, optional summary, optional action items
Storage
Same SwiftData database
Basis
Art. 6 (1) (b) GDPR
Settings
Contents
Language, model tier, microphone, diarisation, auto-cleanup window, crash-report toggle, welcome flag
Storage
UserDefaults on-device
Basis
Art. 6 (1) (b) GDPR
Recording-recovery metadata
Contents
Placeholder row marking a session for recovery if the App was killed mid-recording
Storage
Same SwiftData database
Basis
Art. 6 (1) (f) GDPR — legitimate interest in not losing your recording

Data in this section stays on your device for as long as you keep it. You can delete individual transcripts from the Library, wipe everything at once through Settings → Reset Dikta, or uninstall the App. A timer-based cleanup of audio files is available through Settings → Storage → Auto-cleanup after N days and is off by default.

05
Data Leaving Device

5.1 Subscription Management (Adapty and Apple)

When you open the paywall or check your entitlement, the App communicates with Adapty GmbH (a processor engaged by us under Art. 28 GDPR) to determine whether you are eligible for the free trial, to fetch paywall content, to verify your subscription status, and to process restore-purchase requests. Adapty in turn communicates with Apple's StoreKit.

  • Data sent: an Adapty-generated customer identifier (derived from your device, not from your Apple ID or e-mail), your locale, the product IDs Dikta offers (dikta.premium.monthly, dikta.premium.annual), and the purchase and subscription events Apple reports to the App.
  • Data received: your current subscription status, introductory-offer eligibility, paywall configuration.
  • Legal basis: Art. 6 (1) (b) GDPR — performing the subscription contract.
  • Retention: Adapty retains subscription and entitlement data for as long as we remain their customer or until you request deletion via us.
  • Processor: Adapty GmbH, 140 Broadway, New York, NY 10005, USA (see Section 8 for international transfers).

In-app purchases themselves are processed by Apple, not by us. Apple's privacy policy (apple.com/legal/privacy) governs that processing. We never receive your Apple ID, name, e-mail, or payment details from Apple.

5.2 Crash and Error Reports (Sentry — Opt-Out)

When the App crashes or hits a handled error, and provided you have not switched off crash reporting under Settings → Privacy → Share crash reports, a report is sent to Sentry (hosted in the European Union at ingest.de.sentry.io).

  • Default: crash reporting is on. You can turn it off at any time in Settings, and we will stop sending new reports immediately.
  • Data sent: a stack trace showing which Swift code was executing, the error type and message, your OS version and major device model (e.g. "iPhone 17 Pro, iOS 26.3"), the App version and build number, a sampled subset of internal timing traces, and breadcrumbs describing the last few actions inside the App.
  • Data we actively scrub before sending: any breadcrumb or value whose category or message references the recordings directory, the transcripts directory, your Documents directory, the App's sandbox container path, or an audio-file extension (.wav, .m4a, .mp3, .caf). Screenshots, view hierarchies, and user-interaction traces are disabled.
  • What is therefore not sent: your transcripts, your audio, your titles, your tags, your speaker labels, your settings, your Apple ID, your IP address (Sentry is configured with sendDefaultPII = false).
  • Legal basis: Art. 6 (1) (f) GDPR — legitimate interest in diagnosing crashes and shipping a stable product. You can object at any time (Art. 21 GDPR).
  • Retention: crash reports are retained by Sentry on our behalf for no longer than 90 days, then automatically deleted.
  • Processor: Functional Software, Inc. d/b/a Sentry, 45 Fremont Street, 8th Floor, San Francisco, CA 94105, USA. Sentry's EU ingest region keeps the raw payload inside the EU.

5.3 iCloud Synchronisation (Paid Tier Only)

If — and only if — you have an active Premium subscription and you are signed in to iCloud on the device, the App enables CloudKit-based synchronisation so your transcripts, audio, and paid-tier settings appear across the other devices you use with the same Apple ID.

  • Data synchronised: transcripts (text, segments, words, timestamps), audio files (as CloudKit CKAssets), speaker labels and colours, flags, tags, favourite status, summaries and action items (if present), and a small set of paid-tier preferences. The crash-reporting preference and the "has-seen-welcome" flag are not synchronised.
  • Where it goes: the private CloudKit database of the container iCloud.de.paluro.dikta, hosted by Apple Inc. Data in a private CloudKit database is accessible only to the iCloud account that owns it. We cannot read, list, export, or recover your iCloud data.
  • Legal basis: Art. 6 (1) (b) GDPR — the paid subscription agreement explicitly includes cross-device synchronisation. Apple's own processing of the data in your iCloud account is governed by Apple's Privacy Policy and by the iCloud terms you accepted.
  • Retention: your iCloud data persists until you delete it, either from within Dikta or from outside Dikta (Apple ID → iCloud → Manage Storage). Note: if you turn off iCloud sync after previously using it, your already-uploaded data remains in your iCloud account until you delete it.

5.4 On-Demand Model Downloads

The two smallest Whisper transcription models are bundled with the App and never require a network connection. If you choose a larger model under Settings → Transcription quality, the App downloads it on-demand from the Hugging Face content-delivery network (hf.co and its CDN). The speaker-diarisation component similarly fetches its models on first use.

  • Data sent: a standard HTTPS request — your device's current public IP address (at the network layer, unavoidable for any download), the User-Agent header your OS attaches, and the requested file path. No user identifier, no Dikta-specific header, and no content from the App is transmitted.
  • Legal basis: Art. 6 (1) (f) GDPR — legitimate interest in providing you the model you selected.
  • Retention at our end: none — we do not receive or store this traffic. CDN operators may log standard web-server metadata for short periods under their own policies.
  • Recipient: Hugging Face, Inc., 20 Jay Street, Suite 620, Brooklyn, NY 11201, USA.

5.5 Apple Platform Services (iCloud KVS, Push, App Store)

The App uses a few Apple platform services that process data on Apple's infrastructure under Apple's own terms:

  1. The iCloud Key-Value Store synchronises small preference values across your paid-tier devices when iCloud is signed in and Premium is active.
  2. CloudKit push notifications are used to wake the App when changes land in your private iCloud database. The App does not send any user-visible push notifications on its own.
  3. The App Store and StoreKit 2 process your subscription purchase, renewal, and billing. We receive only the subscription status through Adapty; we do not receive your Apple ID, billing name, or payment information.

All of the above are processed by Apple Inc. under Apple's Privacy Policy (apple.com/legal/privacy).

5.6 On-Device Summarisation (Apple Intelligence)

On compatible devices (iOS 26 / iPadOS 26 / macOS 26 or later, with Apple Intelligence enabled and the required hardware support), the App can produce a short natural-language summary of a transcript. The summariser uses Apple's Foundation Models framework, which runs the base language model entirely on-device on the Apple Neural Engine.

  • Data sent: the transcript text you ask to summarise is passed to Apple's on-device Foundation Models API and processed in local memory on your device.
  • Over the network: nothing. The API we call is an on-device API. We do not transmit the transcript, the prompt, or the generated summary to any server operated by us.
  • Private Cloud Compute: our summarisation call site explicitly targets the on-device model (SystemLanguageModel.default). It does not opt into Apple's optional Private Cloud Compute route. Whether Apple may nevertheless route Apple Intelligence traffic through Private Cloud Compute is governed by Apple's OS behaviour and Apple's Privacy Policy, including Apple's guarantee that PCC does not retain user data.
  • Unavailable: on older OS, ineligible hardware, or with Apple Intelligence switched off, the App reports "Summarisation unavailable" and no processing occurs.
  • Legal basis: Art. 6 (1) (b) GDPR — performing the contract for the feature you invoked.
  • Retention: the generated summary is stored alongside the transcript in the local database (and in iCloud, if Premium + iCloud sync is active). It is never sent to us. Delete it by deleting the transcript.

5.7 What We Never Send

To avoid any ambiguity, the App never sends:

  • your audio recordings to us or to any third party;
  • your transcript text to us or to any third party;
  • your speaker labels, titles, tags, flags, favourites, summaries, or action items to us or to any third party;
  • your microphone input to a cloud transcription service;
  • any advertising or attribution identifier;
  • any cross-app tracking signal;
  • any data to third parties for marketing, profiling, or data-brokerage purposes.
06
Data We Do Not Collect
  • No account data. There is no Dikta account system, so we do not have your name, e-mail address, phone number, or password on any server of ours.
  • No contacts, photos, calendar, health, location, or file-system access beyond the microphone, the specific audio file you explicitly import, and the App's own sandbox.
  • No IDFA, no SKAdNetwork, no third-party SDK that tracks you across apps.
07
Permissions

The App requests the following system permissions at runtime. You can revoke any of them at any time in your device's Settings.

Microphone
Why
To record audio for transcription — core feature permission.
If denied
Recording is disabled; import of pre-existing audio files still works.
iCloud (enabled globally on the device)
Why
To mirror your transcripts and audio into your private iCloud database when Premium is active.
If denied
The App falls back to local-only storage for every transcript.
Apple Intelligence (system setting)
Why
Run optional on-device summarisation of a transcript via Apple's Foundation Models framework.
If denied
Summarisation is disabled; transcription and everything else are unaffected.

The App does not ask for Notifications (it sends none of its own), Photos, Contacts, Calendar, Reminders, Health, HomeKit, Motion, Tracking (IDFA / App Tracking Transparency), or Location.

08
International Transfers

Some of the processors we engage are incorporated in the United States. Where processing under this policy involves a transfer of personal data to a country outside the European Economic Area ("EEA") or the United Kingdom, we rely on appropriate safeguards under Chapter V GDPR, specifically the Standard Contractual Clauses adopted by the European Commission (Decision (EU) 2021/914) and — where applicable — the EU–U.S. Data Privacy Framework adequacy decision of 10 July 2023.

  • Sentry — U.S. entity; crash reports are ingested in the EU (ingest.de.sentry.io). Transfer to the U.S. occurs only for administrative access by Sentry personnel and is covered by the SCCs and Sentry's Data Privacy Framework certification.
  • Adapty — U.S. entity; subscription-state data is processed by Adapty under the Standard Contractual Clauses.
  • Apple — U.S. entity; iCloud, CloudKit, iCloud Key-Value Store, App Store, StoreKit, and CloudKit push notifications are processed by Apple Inc. under Apple's own data-protection regime and the SCCs where applicable. For EU users, Apple processes iCloud data predominantly in European data centres.
  • Hugging Face — U.S. entity; only the public IP and User-Agent associated with a model download are visible to them. Covered by the SCCs and Hugging Face's own data-protection terms.

You have the right to obtain a copy of the relevant Standard Contractual Clauses on request. Please write to [email protected].

09
Legal Bases
On-device core transcription
Art. 6 (1) (b) — Contract
Paid subscriptions (Adapty, StoreKit, iCloud sync)
Art. 6 (1) (b) — Contract
Crash and error reporting (opt-out)
Art. 6 (1) (f) — Legitimate interest in stability
Model downloads from Hugging Face
Art. 6 (1) (f) — Providing selected model
Responding to a privacy request
Art. 6 (1) (c) and (b)
Defending or enforcing legal claims
Art. 6 (1) (f)
10
Retention

We retain personal data only for as long as it is needed for the purposes described above and as long as statutory retention periods require.

  • On your device: for as long as you keep the App installed and do not delete the data. Individual delete, bulk Settings → Reset Dikta, or uninstall at any time.
  • In your iCloud account (Premium): for as long as you keep the data there. Delete it from within the App or from the device's iCloud storage management.
  • At Sentry: 90 days maximum, then automatically deleted.
  • At Adapty: for the life of your subscription plus reasonable periods required for billing and tax record-keeping (typically up to 10 years under German commercial law).
  • At Hugging Face: they do not receive any user-identifying data from the App; only short-lived CDN request logs under their own retention policy.
  • Incoming e-mails to [email protected]: for as long as necessary to respond and for up to three years thereafter for evidentiary purposes, unless longer statutory retention applies.
11
Your Rights (GDPR)

You have the following rights in relation to your personal data:

  • Right of access (Art. 15 GDPR) — obtain confirmation of whether we process personal data concerning you and, if so, a copy.
  • Right to rectification (Art. 16 GDPR) — have inaccurate data corrected.
  • Right to erasure (Art. 17 GDPR — "right to be forgotten") — have your personal data deleted.
  • Right to restriction of processing (Art. 18 GDPR).
  • Right to data portability (Art. 20 GDPR) — receive the data concerning you in a structured, commonly used, machine-readable format. Transcripts can be exported directly from the App as Markdown, plain text, SRT, VTT, PDF, or JSON.
  • Right to object (Art. 21 GDPR) — in particular to processing based on legitimate interest. You can object to crash reporting at any time via Settings → Privacy → Share crash reports.
  • Right to withdraw consent (Art. 7 (3) GDPR) — where processing is based on your consent.
  • Right to lodge a complaint with a supervisory authority (Art. 77 GDPR) — in particular the Landesbeauftragte für Datenschutz und Informationsfreiheit Nordrhein-Westfalen, Kavalleriestr. 2–4, 40213 Düsseldorf, ldi.nrw.de, or the supervisory authority of your usual place of residence.

To exercise any of these rights, write to [email protected]. We will respond within one month (Art. 12 (3) GDPR). We may ask for additional information to verify your identity where the request could otherwise not be honoured safely.

12
How to Delete Everything

You have full control. In order of increasing force:

  1. Delete a single transcript — swipe on the Library row or open the transcript and choose Delete.
  2. Reset the AppSettings → Reset Dikta removes every transcript and audio file on the device; if iCloud sync is active, the deletion is mirrored into your iCloud account on the next sync.
  3. Delete the App — uninstalling the App removes every on-device file in its sandbox. Your iCloud-held copy (if any) is not automatically deleted by uninstalling; see step 4.
  4. Delete the iCloud copy — iOS/iPadOS: Settings → [your name] → iCloud → Manage Account Storage → Dikta → Delete Data from iCloud. macOS: System Settings → [your name] → iCloud → Manage → Dikta → Delete from iCloud.
  5. Ask us to delete everything — write to [email protected]. We will arrange the deletion of any crash reports and subscription records retrievable under our contracts with Sentry and Adapty. Records we are legally obliged to retain (e.g. invoices) will be kept for the minimum statutory period and then deleted.
13
Children

The App is not directed at children under the age of 16. We do not knowingly collect personal data from children under 16. If you are a parent or guardian and become aware that your child has provided us with personal data, please contact us so we can delete it.

14
Security

We rely on Apple's platform security controls (App Sandbox, code-signing, entitlement-based permissions, Keychain-backed iCloud authentication) and on TLS 1.2+ for every outgoing network connection. Audio and transcripts are encrypted at rest on iOS/iPadOS and visionOS by the operating system's data-protection classes, and in iCloud by Apple's at-rest encryption. Despite reasonable technical and organisational measures, no system can be guaranteed absolutely secure; please let us know at [email protected] if you become aware of a vulnerability.

15
Automated Decision-Making

We do not carry out any profiling or automated decision-making within the meaning of Art. 22 GDPR. The transcription and diarisation engines run on your device and produce outputs solely for your own review; they do not produce decisions of legal or similarly significant effect on you.

16
The getdikta.com Website No cookies, no tracking

The marketing website at getdikta.com is a static informational site. We want it to treat visitors the same way the App does — minimally and transparently.

  • No cookies. The site sets no first-party cookies and embeds no third-party cookies.
  • No tracking or analytics. No Google Analytics, no Meta Pixel, no LinkedIn Insight Tag, no PostHog, no self-hosted analytics, no fingerprinting. We do not count visitors at all.
  • No advertising and no cross-site tracking. The site contains no ad networks, retargeting pixels, or third-party marketing scripts.
  • No embedded external content that would make a secondary network request to a different origin (no YouTube iframes, no Google Fonts from googleapis.com, no Gravatar, no social-media share widgets). Fonts and images are served from the same origin.
  • Forms. If you contact us through the e-mail address on the site, your message is processed under Section 11 and retained under Section 10.

Hosting Provider

The website is hosted on Microsoft Azure Static Web Apps in a data-centre region located in Germany, operated by Microsoft Deutschland GmbH, Walter-Gropius-Str. 5, 80807 Munich, Germany (for contractual purposes) and delivered through Microsoft's global infrastructure (Microsoft Ireland Operations Ltd., One Microsoft Place, South County Business Park, Leopardstown, Dublin 18, D18 P521, Ireland, as the EEA contracting entity). In the course of serving the site, Microsoft's infrastructure unavoidably processes technical request metadata — your IP address, the URL you requested, the HTTP status, the User-Agent your browser announces, and a short-lived request identifier — for the purpose of transmitting the page and for security and abuse prevention. These request-level server logs are the only trace any of your visits leaves. They are retained by Microsoft under its own retention policy and are not read, exported, aggregated, or analysed by us.

  • Legal basis for the unavoidable server-log processing: Art. 6 (1) (f) GDPR — legitimate interest in operating, securing, and making available a website that describes the App.
  • International transfers: none of the content on the site is routed to a third country by us. Microsoft may, however, access its own EU infrastructure from outside the EEA for operational purposes; this is covered by Microsoft's Standard Contractual Clauses and the EU–U.S. Data Privacy Framework.
17
Changes to this Policy

As the App evolves, this policy may need to evolve with it — for instance, if we add a new feature that processes data differently, or if a processor's terms change. We will update the Last updated date at the top and, for substantive changes that affect your rights, notify you inside the App before the change takes effect. The current version is always available at getdikta.com/privacy and inside the App under Settings → Legal → Privacy policy.

18
Contact Art. 13 GDPR
Name
Paul Lukas Roder
Address
Montessoristraße 21 40670 Meerbusch Germany
E-Mail
[email protected]

This document is provided in English. A German translation may be offered for convenience; in case of discrepancy, the English version prevails, unless mandatory applicable law requires otherwise.