How to redact a PDF for discovery without uploading it anywhere

A black box drawn over text does not remove the text. If the underlying content stream is still intact, it is one copy-paste or one bad export away from an accidental production. Here is what actually removing content looks like, and how to check that no copy of the file left your machine while you did it.

Written 9 September 2026 · Updated 22 October 2026

Share this guide

Why redact usually doesn't

Most redaction happens by drawing a black rectangle over the part of the page nobody should see. That rectangle is a shape sitting on top of the page content, not a replacement for it. The text underneath is often still present in the PDF's content stream, still selectable, still searchable, and still recoverable by copying the page or by an export step that flattens the annotation layer incorrectly. This is a known, recurring category of production mistake, not a rare edge case: it happens whenever "cover it up" is treated as the same operation as "take it out."

Real redaction means the underlying text and image data for the redacted region no longer exist in the file, not that they are drawn over.

Why uploading it to redact it is itself a disclosure

Before getting to the tool: sending a privileged or confidential file to an online redaction service is a decision, not a neutral step. That file leaves your machine and your firm's control the moment it is uploaded, whether or not the vendor "just processes and deletes it." For discovery and disclosure work specifically, that is often a fact that would need to be disclosed and justified if anyone asked, and it rarely gets treated as one.

The alternative is software that never uploads the file at all, which removes the question rather than requiring you to trust an answer to it.

The workflow

1. See what's actually in the file first. X-RAY inspects a PDF's structure and content before you touch it: embedded metadata, hidden layers, prior revision history some PDFs carry without anyone intending it.
2. Redact for real. SEAL removes the marked region's underlying content, not just its visible appearance, and produces a file where the redacted text is genuinely gone. It is one file at a time with fully manual marking: no batch mode, no keyword auto-detection. For a handful of documents that's a feature (you see exactly what you're removing); for a large-volume production run it is a real limit, stated here rather than glossed over.
3. Keep a record of where it came from. CHAIN and PROVENANCE add the chain-of-custody notes that disclosure work usually needs to show later: what the file was, when it was processed, and what was done to it.
4. Produce it. DOCKET and EXHIBIT handle the assembly and numbering for what actually goes out the door.

None of these upload anything. Here is how to check that yourself rather than take it on faith: open your browser's network tab (F12, Network), run a real file through SEAL, and watch nothing get sent anywhere. Then disconnect from the internet entirely and confirm it still works. Software that is quietly phoning home cannot do that. Full writeup, including what the browser itself enforces regardless of what this page says, at /verify.

What this is not

Not a document management system. No bulk processing. No multi-user review. No claim that a bundle of output satisfies any particular court's specific redaction rules; check those yourself for your jurisdiction. This is a set of tools for the disclosure and evidence workflow, not a replacement for legal judgment about what needs to come out of a document.

Cost

Checking a PDF is free, with no account. Saving the redacted file is part of OBSCURA Pro, or US$2 to save from one tool for a day. The rest of the suite is at /record.

Part of the Record Suite. Every tool linked here is free to open and to check a file with, with no account, and never watermarks what it gives you. Saving the result is part of OBSCURA Pro.