Redact PDF Pages by Search Term
Auto Redact replaces every page whose text contains one of your search terms with a solid black page. Matching ignores case, and the old content of a redacted page is removed. It works on text layers, not scans.
Drag and drop files here, or click to choose
Accepts PDF · up to 500.0 MB per file · up to 20 files at once
You can also paste a file with Ctrl/Cmd+V, or from the clipboard menu.
Auto Redact reads the text of each page and compares it with your comma-separated terms. Matching is a case-insensitive substring test, so `acme` also hits `ACME Corp`. Every page that matches is replaced by a solid black page. It is page-level redaction, not word-level.
On a matching page the old content, resources and annotations are detached, so the hidden words cannot be extracted from that page. The fill is fixed at black (`#000000`); the color field is ignored.
The tool also scrubs what a redacted page leaves elsewhere in the file: form fields on it, bookmarks that point at it or contain a term, matching document-information entries, XMP metadata of redacted pages, and structure-tree text. Embedded files, JavaScript and text on pages that did not match are not touched; Sanitize removes the first two.
An empty term list is rejected with an error. Pages with no match, including image-only scans that have no text layer, come out unchanged. This tool does no OCR: run OCR PDF with the Searchable PDF output first so there is text to match, or remove the scanned page with Remove Pages.
The API also accepts `caseSensitive=true` and `useRegex=true` (Rust regex syntax, at most 200 comma-separated terms). The website form has only the term list.
The page-level design is the single most important thing to understand, because it makes Auto Redact very different from the black-box redaction people picture. You do not draw rectangles over a name; you list terms, and any page containing one is replaced wholesale by a solid black surface. That is deliberate: word-level redaction is where most real-world leaks happen, because a black box drawn over text can often be removed or the underlying text can still be selected and copied. Removing the entire page's content makes the words genuinely unrecoverable from that page. The trade-off is that you lose everything else on the page too.
That means the right workflow starts with page selection, not with the term list. If the sensitive value appears on page 7 of a 40-page report, you have to decide whether losing page 7 entirely is acceptable. When it is not, the common pattern is to extract the sensitive pages into their own file, redact them there, and distribute a packet that references the redacted pages without including the surrounding content. Because redaction is destructive, always work on a copy and keep the original under your own retention rules.
Matching is a plain substring test, so short or ambiguous terms cause over-matching. A term like `id` will black out every page that contains the word `identity`, `valid`, or `provide`. Prefer the longest unambiguous string you can - a full name, a complete account number, a unique reference - and inspect which pages went black before sending the result. The API's regex mode exists for exactly this reason: an anchored pattern like an account-number shape can be far more precise than a bare substring.
Password-protected files are unlocked for you on this page once you enter the password; API callers run Unlock first.
Features
- Case-insensitive term matching; every matching page becomes a black page
- Removes the page's old content, form fields, matching bookmark titles and metadata
- Empty term list is rejected; pages with no match are unchanged
- Page-level and destructive; text is not merely covered
- Anonymous API at /api/v1/security/auto-redact
When to use this tool
- Black out pages that mention a client name or account number before sharing
- Remove pages that contain personal identifiers from a long export
- Prepare a review copy in which sensitive pages must not be recoverable
- Produce a version of a filing where specific referenced pages are withheld
How do I redact pages in a PDF?
- Upload a PDF with a text layer that you are authorized to alter.
- Enter comma-separated terms, for example a client name or an account number. Matching ignores case.
- Click Process. Every page that contains a term becomes a solid black page.
- Open the result, check which pages went black, and search it once more for the terms before you share it.
Limits and edge cases
- Page-level only; there is no partial redaction on a page
- Needs a text layer: image-only pages never match
- An empty term list is an error
- Embedded files and JavaScript are not removed; use Sanitize for those
- Words drawn as vector shapes have no text to match
- Short terms over-match; the whole page is lost
- Upload limit on this website: 500 MB per file, sent in chunks above about 95 MB. A single direct API request body is capped at 100 MB.
Examples
- The term secret blacks out page 2 only when that page's extracted text contains secret in any capitalization
- The terms alpha,beta black out every page that contains either word
- The term id over-matches identity and valid, so a longer term is safer
Privacy for this tool
Files are uploaded over HTTPS, processed in memory or a short-lived temporary directory on our servers, and deleted when your result is ready. We do not keep copies for later browsing, training, or advertising profiles. See the Privacy Policy for retention details and AdSense cookie disclosures.
Frequently asked questions
- Why did the whole page go black?
- Redaction is page-level. Any page whose text contains a term is replaced by a black page, so nothing on it survives.
- Can I redact one word and keep the rest of the page?
- No. The whole matching page is blacked out; there is no word-level box redaction.
- Is the text really removed from the file?
- On matching pages the content, resources and annotations are replaced, so the text is no longer in the page. Form values, matching bookmark titles, Info entries and metadata are scrubbed as well. Embedded files and pages that did not match are left alone.
- Is matching case-sensitive?
- No. By default terms match regardless of capitalization. API callers can send caseSensitive=true, or useRegex=true for regular expressions.
- Can Auto Redact handle a scanned document?
- Not directly. It reads a text layer, and a scan without one is left unchanged. Run OCR PDF with the Searchable PDF output first to give it text, then check every page that should have gone black, because OCR can misread a term.
- Why did more pages go black than I expected?
- Matching is a substring test, so a short term like `id` also matches `identity` or `valid`. Use the longest unambiguous term you can, or a regex through the API, and check which pages changed.
- Can I recover a page after redacting it?
- No. Redaction is destructive. Run it on a copy; the original on your own disk is unchanged, but the redacted file cannot be reversed here.
Last updated:
Call this from code
Every tool on this site is a plain REST endpoint - no account or API key needed for anonymous use. Built for AI agents and developers as much as for browsers.
curl -X POST "https://pdf123.xyz/api/v1/security/auto-redact" \
-F "[email protected]" \
-F "listOfText=your-listOfText" \
-o output.pdfAlso available as an MCP tool for agent clients that speak Model Context Protocol (JSON-RPC 2.0 over POST /mcp). Full API reference
Use it from an AI agent
SkillClaude Code, Codex, Cursor and other AI agents can run Redact for you with this skill: /skills/pdf123-auto-redact.md
Files are used only for this processing job and deleted automatically afterward.