Redact PDF Online Free: Remove Text Permanently
Drag boxes, search for words, or scan for SSNs, card digits, emails, and phone numbers, then flatten the marked pages so the text is gone from the file, not just painted over.
or click to browse
How to Redact a PDF So the Text Is Actually Gone
Most redaction tools fail at the last step. Here the pipeline is built to survive copying, searching, and extraction:
- Mark the content - Drag boxes by hand, search for a word or phrase to mark every hit, or run the PII scan for Social Security numbers, card digits, emails, and phone numbers. Review the boxes before anything is applied.
- Apply with flatten mode - Every marked page is re-rendered through pdf.js at 150 or 300 dpi with the boxes painted in, then rebuilt as a single image page. The old content stream, with its text operators, is not part of the output anymore.
- Scrub the metadata - Document info fields are cleared, date stamps reset, and the XMP packet is detached from the catalog and blanked at the byte level so editor history does not survive either.
- Read the verification report - Text runs per redacted page are counted in the output itself: zero means zero. Metadata fields are shown before and after, and the report ends with a copy-paste test you can run in any viewer.
Flatten vs Overlay, and What Each One Proves
The difference decides whether your document is safe to send:
- Flatten (default) - The page becomes an image. Text can never be selected, searched, copied, or extracted from under the box because no text exists there. The cost is that the redacted page loses its text layer, which is the point.
- Overlay - Black rectangles drawn on top look redacted but keep the original text inside the file. It exists for quick visual drafts and is labeled not secure in the tool for a reason.
- PII pattern pack - SSNs, credit card style digit runs, emails, and phone numbers are matched with regular expressions across every page, plus a custom pattern field for case numbers, project codes, or anything specific to your document.
- Metadata and XMP - Title, author, subject, keywords, creator, producer, dates, and the XMP packet are the quiet leaks that survive a visual redaction. The scrub checklist handles all three groups and reports each field.
- Untouched pages - Pages without boxes are copied as-is, so contracts and reports keep their searchable text everywhere except where you redacted.
Redact Documents Without Uploading Them
Redaction content is exactly the material you least want on a third-party server: legal filings, medical records, financial statements, identity documents. This page never asks for an upload. pdf.js runs the text scan and the flatten render, pdf-lib assembles the new document, and the verification pass reads the finished bytes back to confirm the counts in the report are real. After redacting, check Unlock PDF if the source was protected, clean structure with Repair PDF, or finish the document with Edit PDF. For site-wide checks, run the Free SEO Audit Tool or book our SEO audit services.
Frequently Asked Questions
How do I redact a PDF?
Drop the file, then either drag boxes over the content to hide, use Find and mark to target exact words, or run the PII scan for emails, phone numbers, SSNs, and card digits. Press Apply Redactions and download the result.
Is the removed text really gone from the file?
In the default flatten mode, yes. Each marked page is re-rendered to a lossless image with the boxes baked in, so the original text operators no longer exist in that page. The report proves it by counting text runs on redacted pages in the output: zero. The optional overlay mode only paints black rectangles and is labeled not secure for a reason.
Can it find sensitive information automatically?
Yes. The PII pack scans the text layer of every page for US Social Security numbers, credit card style digit runs, email addresses, and phone numbers, and a custom regular expression lets you match anything else. Every hit becomes a redaction box you can review before applying.
What metadata does the scrub remove?
Title, author, subject, keywords, creator, and producer are cleared, creation and modification dates are reset, and the XMP metadata packet is deleted from the catalog and blanked out of the file bytes. The report lists each field with its before and after value.
What does the verification report show?
Mode used, pages and box counts, text runs before and after on every redacted page measured from the actual output, metadata fields before and after, XMP status, and a copy-paste test to run yourself in any viewer.
Is my file uploaded?
No. The pdf.js text scan, the canvas flattening, the pdf-lib rewrite, and the metadata checks all run inside this tab. The redacted copy downloads directly from your browser.
What happens to pages I did not mark?
They are copied through untouched and stay fully searchable and selectable. That is deliberate: only the pages you marked are flattened, so the rest of the document keeps its text layer.