You redacted the document. You're about to release it. Before you do, run the 30-second check that catches almost every redaction failure.
Step 1: select-all and copy
Open the redacted PDF in any reader (Acrobat, Preview, browser). Press Cmd-A / Ctrl-A on each redacted page to select all text. Press Cmd-C / Ctrl-C to copy.
Paste into a text editor. Look at the redacted regions. If nothing appears where the redactions are, the underlying text has been removed. If the redacted text appears in the paste, redaction was visual only — the document is leaking.
Step 2: search for known terms
Use the PDF reader's find function to search for the terms you redacted. If a search for 'John Smith' returns hits in the redacted PDF, the text is still in the file. If search returns no hits, the text is properly gone.
This catches cases where redaction was applied to some instances but missed others — a common error when redacting manually across many pages.
Step 3: inspect metadata
Open the PDF's document properties (File → Properties in most viewers). Check:
- Title: doesn't contain the redacted terms. - Author: doesn't reveal who created or last edited. - Keywords: doesn't list redacted content. - Comments: empty or stripped.
Strip metadata if any of these contain sensitive material.
Step 4: examine the structure
For very sensitive releases, use a PDF structure inspector (pdfinfo, qpdf, PDF Stream Dumper) to confirm:
- No incremental updates that contain the original (un-redacted) content. - No embedded files or attached objects with sensitive content. - Cross-reference table consistent with the rendered output.
This is forensic-grade; overkill for routine sharing but appropriate for high-stakes public releases.
FAQ
What's the most common failure?
Visual-only redaction — black rectangles drawn over text. The select-and-copy test catches almost all of these instantly.
Should I always strip metadata?
For external sharing, yes. Metadata leaks have caused several public-release incidents. Strip as part of standard release workflow.
Can OCR re-reveal redacted text?
If the underlying text wasn't removed (visual-only redaction), yes. If true redaction was applied, OCR has nothing to read — the bytes are gone.
Is there a tool that automates verification?
Some legal-tech and security tools include redaction QA. For most workflows, the manual select-and-copy test is fast and reliable.
Thirty seconds of verification beats a redaction failure in the news. Redact in Flint, test, then release.