Privacy

Your pixels are not our business

Last updated: August 28, 2026.

Local processing

Selected images, PDF bytes, recognized text, and editable results are processed in your browser. PDF passwords are held only in the current document worker's memory and are not saved. BrowserOCR does not provide an upload endpoint, cloud OCR fallback, account system, or ads.

Anonymous measurement

Cloudflare Web Analytics counts page visits. Beginning September 5, 2026, BrowserOCR will also count three anonymous task steps: a task started, a usable result completed, and a result copied or downloaded. Each step is counted at most once per page load. These records contain only the site hostname, public page path without query parameters, a coarse source category, the step name, and time.

Network requests

Your browser requests application code, same-origin PDF/WASM resources, ONNX Runtime files, and OCR models from browserocr.com. High accuracy model files are requested only after your confirmation. The anonymous task endpoint never receives image or PDF bytes, filenames, file types, sizes, passwords, recognized text, editable results, complete referrers, query parameters, error details, cookies, or user identifiers.

Device storage

A service worker caches the application, PDF engine, and model resources so they can work offline after loading. Your selected model preference is stored in local storage. The documents and results themselves are not cached by BrowserOCR. Clear this site's browser data to remove the resource cache and preference.

Contact

For privacy questions or support, email [email protected].