Privacy · Architecture
Free OCR without uploading your file.
Most free OCR sites ask you to upload a document, wait in a queue, and trust that the file is deleted afterwards. This site takes a different route: the OCR engine runs in your browser.
The difference
Upload-and-wait versus run-in-place.
A hosted OCR service needs a copy of your document before it can read it. You select a file, the browser sends it to a server, software on that server performs recognition, and the text comes back. The service may delete the upload quickly, but the document has still left your device.
On this site, the document is loaded into the memory of your browser tab. PDF.js decodes the pages, and Tesseract.js — an OCR engine compiled to WebAssembly — recognises the characters. The extracted text is written back into the page you are already looking at. There is no upload step to delete because there was no upload.
The trade-off is that your device does the work. A modern laptop handles a few dozen pages comfortably. A phone can process a handful, but very large files may hit memory limits. For occasional documents, that is usually a good exchange for not sending the file anywhere.
1. The page loads
HTML, CSS, fonts and the OCR widget are downloaded. The engine and language data follow when you first use a tool.
2. Your file stays put
The file picker gives the page temporary access to the file. PDF.js and Tesseract.js read it in the browser tab.
3. Text appears locally
Recognition output is placed in the workspace. Copy it or export it; close the tab and the page contents are gone.
What still travels over the network?
Being precise matters more than a slogan. This site does make network requests; it just does not send your document.
| Item | Sent to a server? |
|---|---|
| Your PDF or image | No |
| Extracted text | No |
| OCR engine and language data | Downloaded from a CDN |
| Page and font assets | Downloaded from this site |
| Analytics | Only after you accept cookies |
Analytics is off until you choose “Accept all cookies” in the cookie banner. Rejecting non-essential cookies does not disable any tool. See theprivacy policyfor the full description.
Documents that should not be uploaded casually.
Local OCR is useful whenever the content is sensitive or the retention policy of a hosted service is unclear. Common examples:
- Identity documents such as passports, licences and national ID cards.
- Medical letters, prescriptions and clinical notes.
- Bank and credit card statements, payslips and tax forms.
- Contracts, NDAs and internal company reports.
- Student records or anything containing another person's personal data.
None of this means a hosted OCR service is unsafe by definition. It means the decision depends on the provider's retention and security practices. Local processing removes that dependency for the document itself.
No-upload OCR questions.
More free tools
Private, browser-based tools that run on your device.