PDF redaction for firms: irreversible removal, not black boxes

Black boxes look secure and often are not. Many tools only place a shape over visible text - while characters remain in the text layer, in embedded images or in earlier revisions. For firms sending papers to courts, agencies or opponents, that is an avoidable risk.

Covering is not removing

True redaction removes content from the file structure. Afterwards, neither search nor copy should restore the redacted passage. A simple test after export: search for a unique redacted term, use a select tool over the area, run OCR again on the page. If any of those recover something, the process was incomplete.

Scanned pages are especially risky when redaction sits only in the image while an older OCR layer still holds clear text. So are form fields, comments, embedded files and attachments inside the PDF.

What typically needs redaction

Beyond classic personal data (account numbers, private addresses, health data), practice often includes internal notes, strategy remarks and third-party data not required for the proceedings. The aim is not maximum redaction but appropriate redaction: explainable, justified, consistent across the bundle.

Release checklist

• Define categories to redact for the matter (firm rule).
• Keep working copy and original strictly separate.
• After redaction: search, select and zoom spot checks.
• Check metadata and document title (author, firm paths).
• Avoid careless re-OCR after redaction; rebuild deliberately if needed.
• Second-person release on sensitive packs.

Process before tool

Features help only with clear ownership. Assistants often propose marks; counsel confirms categories and samples. Time pressure tempts people to “draw boxes quickly”. That is when gaps appear. Better: mark critical passages, redact in one pass, then verify.

LexLogik supports document workflows including quality checks in the browser; what to redact remains a professional decision. Where true redaction is unavailable in the tool in use, change tool or regenerate the page as a new image without a text layer - and record that choice.

Technical evidence of redaction

Note in the file: tool, date, release, categories removed. That saves explanation when scope is disputed. Use “print as image” only when true redaction is missing - and record why. Train cover staff expressly: under pressure redaction is the step done “approximately”.

Text layers and older OCR

Scanned pages with an old OCR layer under new image blackout are especially risky. After redaction, test search, select tool and recognition again if needed. Form fields, comments and embedded attachments belong in the same check.

Traceability

When the scope of redaction is later disputed, a short internal note on why categories were removed saves explanation to client and court.

Check the text layer after redaction

Open the redacted file in a viewer with text selection and try to select under the bar. If clear text copies, the redaction was only visual. Save only the cleaned export into the filing file.

Redaction protects only when characters and metadata are truly removed - not when a black box sits on readable text. Combine irreversible redaction, text-selection spot checks and a clear release step, and you avoid embarrassing clear-text finds at court or with the other side.