Your document has a short visit
No saved uploads or results
The OCR application processes uploaded bytes, rendered pages and recognized text in server memory. It does not save these to files, databases, object storage or request logs. Your browser keeps its active document and result state in memory; it does not put these in cookies, local storage, IndexedDB or a service worker.
Ten minutes, including processing
The server starts the 10-minute deadline when upload begins. Expiry removes the session and stops its worker, including unfinished extraction. Delete session now removes it sooner. Restarting the server also discards all sessions. Closing a page clears its browser state; server expiry remains enforced independently.
Private access to results
A random session identifier and separate bearer token control access. Keep those credentials private. They are not placed in result links. Downloaded text files are saved only when you request them and remain on your own device until you remove them.
Basic abuse prevention
The service briefly retains a salted network-address hash and request count in memory to limit excessive uploads. It does not retain the raw address in that limiter or log upload names, document contents or recognized text. Hosting operators must keep proxy body buffering, access logging, swap and crash dumps disabled as described in the deployment guide.
Ads and public pages
The default build has advertising disabled and contains no analytics trackers. OCR workspaces never load advertising scripts. If advertising is enabled later on public information pages, this policy must be updated to identify the active providers, consent choices and their data use before launch.
Use your judgment
OCR is not suitable for every sensitive document. This application cannot control device-level browser extensions, downloads you save yourself or an incorrectly configured hosting platform. Do not upload material you are not entitled to process.