To redact a PDF for court filing, you must destroy the sensitive content, not cover it — and you should do it without uploading a privileged document to someone's server. The Redact PDF tool does both: you draw boxes over sensitive content, the tool rebuilds those pages with the content permanently removed, and the entire process runs in your browser. Free, no account, nothing transmitted.
What the rules actually require
Federal Rule of Civil Procedure 5.2 sets the baseline for federal filings. Unless the court orders otherwise, filings may include only:
- the last four digits of a Social Security or taxpayer-identification number
- the year of an individual's birth date
- a minor's initials instead of their name
- the last four digits of a financial account number
Criminal and bankruptcy rules have parallel provisions, and many districts layer local requirements on top: some mandate flattened PDFs, some require certification that metadata has been removed, and several publish explicit warnings about failed redactions. The rule that matters is the one in your judge's standing order — check it. But every variant shares the same premise: the information must be absent from the file, not merely hidden.
The redaction failure hall of fame
The legal profession relearns this lesson on a regular schedule. In widely reported cases, attorneys have "redacted" filings by drawing black rectangles over text or setting highlight color to black — and journalists recovered the hidden content within minutes by selecting the text and pasting it elsewhere. The consequences range from professional embarrassment to disclosure of sealed material, privilege waiver arguments, and bar discipline.
The mechanics of the failure are always the same: a PDF is a layered format. A rectangle drawn on top of text is just another layer. The text object underneath remains in the file, fully intact, one copy-paste away from disclosure. The same applies to:
- Highlighting in black — the text is under the highlight.
- White text on white background — still selectable.
- Cropping pages — cropped content is hidden, not deleted.
- Image stamps over text — the text layer survives beneath.
How to redact properly, step by step
- Identify everything FRCP 5.2 (or your local rule) covers. Search the document systematically: SSNs, account numbers, birth dates, minors' names. Don't forget exhibits, headers, and footers — and remember that scanned exhibits can contain handwritten identifiers.
- Open the Redact PDF tool. Drop your document in. It loads into your browser only; there's no upload step because there's no server.
- Draw redaction boxes over each item, page by page. Cover the full extent of the text, with margin.
- Apply redaction. The tool re-renders each affected page and rebuilds it as a flattened image with the boxed regions destroyed first. The output file does not contain the redacted text in any layer.
- Verify your work. Open the output, try to select text where the redactions are, and run Ctrl+F searches for the redacted strings. Nothing should be findable. This sixty-second check is the difference between redacting and hoping.
- Strip the metadata. Run the output through the Metadata Viewer to inspect and remove author, software, and timestamp fields before filing.
Why local processing matters for legal documents
Uploading an unredacted filing to a third-party web service — the standard workflow of most online redaction tools — means transmitting the very information you're obligated to protect to an unknown server, before the redaction has happened. Whatever that service's deletion policy says, you've created a disclosure you can't audit and can't take back.
In-browser redaction removes the transmission entirely. The document is processed on your machine by open-source engines; you can watch the network tab show zero requests, or work offline. For privileged and sealed material, that's not a feature preference — it's the only architecture consistent with your obligations.