What happens when you upload to a typical OCR site
The workflow of nearly every popular online OCR tool looks like this:
- You upload your image.
- The image travels over the internet to the tool's server.
- The server runs OCR software on it.
- The extracted text is sent back to your browser.
- Your image — and possibly the extracted text — may be stored on their server, sometimes for days or permanently.
Most privacy policies for these tools contain language like "we may retain uploaded content for service improvement." For a photo of a birthday cake, that's fine. For a photo of your passport, tax return, bank statement, or private message — it's a serious risk.
The privacy-safe alternative: browser-based OCR
WebAssembly (WASM) has changed what's possible in a browser. It allows complex software that previously required a server — compilers, video editors, image processors — to run entirely inside your browser tab. Tesseract.js is a full OCR engine compiled to WebAssembly.
When you use imagetotextocr-phi.vercel.app, this is what actually happens:
- You select an image from your device.
- The file is read by JavaScript running in your browser tab (not sent anywhere).
- Tesseract.js — already loaded in your browser — processes the image locally.
- The text is displayed in your browser.
- No image data ever leaves your device.
How to verify this yourself
You don't have to take our word for it. Open the tool, then open your browser's Developer Tools (right-click → Inspect → Network tab). Select an image and click Extract Text. Watch the Network tab — you'll see the Tesseract language file load (that's normal, it's the OCR model), but you will see zero requests that upload your image data.
Try it — your documents stay on your device
Open the private OCR tool →When does this matter most?
- Passport and ID photos — OCR on travel documents is common for visa applications. Never upload these to unknown servers.
- Tax documents and financial records — receipts, invoices, bank statements contain personal financial data.
- Medical documents — prescriptions, test results, insurance letters.
- Personal messages — screenshots of private conversations.
- Business documents — contracts, NDAs, internal reports.
The open source transparency guarantee
This tool is fully open source. The source code is available on GitHub for anyone to audit. The privacy guarantee isn't a claim in a terms of service — it's verifiable in the code itself. There are no upload calls. Period.