User Disclosure

Updated October 17, 2026 · Vision OCR disclosure v1

Read this before processing a document. These disclosures explain the service; they do not remove rights you have under applicable law.

Check the source before relying on a result

OCR can misread faint scans, handwriting, columns, tables, punctuation, and numbers. AI can return incorrect or unsupported fields. Compare the output with the original. DocuExtracts does not verify authenticity, signatures, legal validity, or whether an invoice is payable.

Use files you are authorized to process

You must have permission or a lawful basis to process documents and personal information belonging to others. Redact unnecessary personal data first. Do not submit passwords, authentication secrets, full payment-card details, or regulated health records. The service offers no certification or special compliance agreement for sensitive or regulated workloads.

Know what leaves your device

Local extraction reads original files in your browser. Vision OCR instead sends rendered document page images through our server to OpenAI after you explicitly allow it. Images are not saved by DocuExtracts; provider retention policies apply. AI field extraction sends extracted text to OpenAI. Selecting Save stores extracted text, fields, selected outputs and extraction warnings. For signed-in local processing, filenames, page counts, status, and usage are recorded on the server; vision requests record AI usage. Guest local results are lost when you close or refresh the page unless downloaded or saved after signing in. Hosting providers record technical logs, and the operator receives activity summaries. See the Privacy Policy for details.

Vision OCR limits

Vision OCR requires sign-in. Each successfully processed page uses one AI action: five free per calendar month shared with AI field extraction, then one AI credit per page. Local processing page credits are separate. Failed page requests release their reserved AI allowance when the database is available. Earlier successful pages still use allowance if a later PDF page fails; the result identifies that partial coverage. Vision OCR does not guarantee table geometry, numeric accuracy, or complete transcription. Merged-cell text appears in the top-left cell, with covered cells left empty. Review warnings and every result against the original.

Understand limits and credits

Guests can extract and download without signing in. Signed-in accounts receive five free processing pages; additional account processing requires page credits. PDFs longer than five pages are limited to their first five pages, with a notice in the upload flow. The remaining pages are not extracted. Other supported files count as one processing page. A credit reservation or successful HTTP response alone does not establish that extracted information is accurate or complete.

Keep your own copies

Keep original documents and export important results. Unsaved results may disappear when you close or reload the page. Service interruptions can occur. Contact gayasi007@gmail.com about account, processing, privacy, or billing concerns.

Questions? Contact the DocuExtracts site operator at gayasi007@gmail.com.