InstaScan AI
Privacy Policy
InstaScan AI · Last updated 13 October 2026
InstaScan AI is a free tool that reads text from images and PDFs. All recognition happens in your web browser. Your files are never uploaded to us or to anyone else. The only time text leaves your device is when you choose online translation (see below).
Files you add
Images and PDFs you open, drop, paste or photograph are processed in your browser's memory. They are not sent to any server and are gone when you close or reload the page.
Google Drive
If you choose Google Drive, you sign in with Google, which asks you to grant the drive.file permission. That permission lets the app open only the files you pick in the Google file picker, nothing else in your Drive. Those files download directly from Google's Drive API into your browser.
- We don't store, copy, share or sell your Google data.
- We don't modify or delete files in your Drive.
- The sign-in token stays in your browser's memory only for the session.
- You can remove the app's access at any time at myaccount.google.com/permissions.
The app's use of information received from Google APIs follows the Google API Services User Data Policy, including the Limited Use requirements.
Translation
The Translate tile translates recognized text between English and Indian languages. Only the text is involved, never your image or PDF.
- On this device: in browsers with a built-in translator (current Chrome and Edge on desktop), translation runs locally and the text stays on your device. The browser downloads each language pair once.
- Online, only if you choose it: in other browsers, or for language pairs your browser can't handle, the button reads Translate online. Clicking it sends the recognized text to the free MyMemory service run by Translated srl, under its own terms. Don't use online translation for confidential text.
Search by image and by description
Image search (duplicates, and finding a picture or a piece of one inside your images) and Describe search (finding images by what they show) both run entirely in your browser. Your images and your query are never uploaded.
- Image search uses the open-source OpenCV library, downloaded from jsDelivr the first time you use it.
- Describe search uses Apple's MobileCLIP model. After you agree, your browser downloads it (about 88 MB) from Hugging Face once and keeps it for next time. Only the model files are downloaded; nothing about your images is sent.
What your browser stores
Your recognition settings, translation languages, palette and light/dark theme choice are saved in your browser's local storage. Downloaded language models are cached by your browser so the next run starts faster. You can clear both by clearing this site's data in your browser.
Third-party services
To run, the page loads code and fonts from these services, which may see your IP address as part of a normal web request:
- jsDelivr: the OCR engine, language models and PDF reader
- Google Fonts: the page's fonts
- Google Identity Services and Google Picker: only when you use the Google Drive button
- MyMemory (translated.net): only when you click Translate online
- Hugging Face: only to download the Describe search model, after you agree
- Vercel or GitHub Pages: hosting for the page itself
The app has no analytics, advertising or tracking cookies.
Contact
For questions about this policy, open an issue on the project's GitHub repository.