Privacy Policy

Last updated: 2026-11

For Apple Platforms (iOS / iPadOS / macOS)

This page explains OCREditor's data flow on iPhone, iPad, and Mac. Packaged Apple apps and native workflows use on-device Apple Vision / CoreML OCR by default; if you deliberately switch to Gemini cloud OCR, only the selected image region is sent to Google's Gemini API.

1. Processing Modes and Data Flow

  • Local by default: native OCR, browser Tesseract OCR, background repair, text editing, formatting, erase-region, copy/paste region, and exports run on your device.
  • Cloud OCR is optional: image data is sent to Gemini only when you actively choose Gemini OCR, provide your own API key, and run re-recognition or regional OCR. That processing is governed by Google's terms and privacy policy.
  • No advertising telemetry: the documentation site and editor do not bundle ad SDKs, and edited content is never proactively uploaded to our servers.
  • Apple on-device OCR: packaged iOS/iPadOS/macOS apps call Apple Vision directly on your device; the macOS desktop browser workflow can also connect to your local OCR service on 127.0.0.1:5001 running Apple Vision on your own Mac.

2. Local Storage, Cache, and Temporary Data

To keep the editor working smoothly, the browser or app may store the following locally:

  • Settings and preferences: UI language, OCR engine choices, font preferences, word fine-tuning mode, auto-OCR toggles, and your optional Gemini API key are stored solely in local browser or app storage.
  • Edit history: undo and redo states are kept in the current session's memory to preserve canvas actions and are cleared when the session ends.
  • Models and caches: if you enable high-quality AI background repair (LaMa), the model is downloaded once and cached locally on your device for reuse.
  • Export and file control: Save Image and Save Image As write only to the location you choose or trigger a local download.

3. Permissions and Platform Behavior

Common Apple platform (iOS / iPadOS / macOS) permissions and behaviors include:

  • Photos or file access: used only to open images you pick and to save exported results.
  • Camera permission: if you import by taking a photo, the captured image is processed locally on your device and is never uploaded to us.
  • Fonts and system resources: the app uses fonts available on your Apple device (and bundled Noto Sans/Serif TC) for rendering and does not upload your font list.
  • Hardware acceleration: OCR, background repair, and canvas rendering may use the GPU or Apple system AI frameworks (CoreML / Metal / WebGPU), without accessing unrelated files.
  • macOS localhost OCR service: when using the desktop browser with macOS native OCR, the browser communicates with a local service on the same Mac rather than an external server.

4. Controls You Can Use to Limit Network Activity

  • Avoid external services completely: keep using Native OCR or browser Tesseract OCR and leave the Gemini API key empty.
  • Disable automatic OCR: load only the source image first, then trigger recognition manually when you are ready.
  • Exports always finish locally: even if you used optional cloud OCR earlier, the final edited canvas and exported image are generated on your device.

5. Policy Updates

If features or data flows change materially, we will update this page. Enabling cloud OCR always remains your explicit choice.

Last Updated: November 2026