A PDF looks like a sheet of paper, but it behaves like a small database. Behind the visible page sits a text layer, an object tree, metadata streams, and sometimes scripts and embedded files. Most PDF "leaks" happen not because someone cracked encryption, but because the sender misunderstood one of those layers. These five tips cover the mistakes that actually cause incidents — and the fixes that actually work.

1. Understand what a PDF password actually does

The PDF specification defines two different passwords, and confusing them is probably the single most common PDF security mistake.

• The user password (or "open password") encrypts the document's contents. Without it, the file is ciphertext — nobody can read the pages.
• The owner password (or "permissions password") does something much weaker: it sets flags that politely ask PDF viewers to disable printing, copying, or editing.

Here is the uncomfortable part: owner-password restrictions are advisory. When only an owner password is set, the file is encrypted with a key the viewer can derive without knowing any password at all — it has to be, or the document couldn't be displayed. A compliant reader like Adobe Acrobat honors the no-print and no-copy flags, but dozens of open-source libraries simply ignore them, and "PDF password remover" tools strip them in about a second. If you have ever received a contract you "couldn't copy text from," you have seen this theater in action.

Real protection means a user password with modern encryption. Choose AES-256 (available since PDF 2.0 and in most tools for years) rather than the legacy RC4 schemes, which are considered broken. And because the encryption key is derived from your passphrase, the passphrase is the weakest link: a short word falls to offline guessing regardless of the algorithm. Use a long passphrase — four or five random words — and send it through a different channel than the file itself. A password emailed in the same thread as the attachment protects against essentially nothing.

2. Redact properly — black boxes are not redaction

Drawing a black rectangle over text hides it from your eyes and nobody else's. The text still exists in the file's content stream underneath the shape. Anyone can select the "hidden" region and copy it, run the file through a text extractor, or open it in an editor and delete the rectangle. Court filings and government releases have been embarrassed by exactly this failure repeatedly — it is a recurring genre of news story for a reason.

True redaction is destructive: it removes the text and image data from the content stream and replaces it with a placeholder, so there is nothing left to recover. Dedicated redaction features in professional PDF editors do this. Highlighter tools, comment boxes, and shape annotations do not.

When an entire page is sensitive — a salary schedule in a board packet, an appendix of account numbers — the safest move is also the simplest: remove the page entirely with a delete pages tool before sharing. Deleting a page removes its content objects wholesale, with no risk of a stray text run surviving under a rectangle. It is much harder to get wrong than selective redaction.

3. Strip the metadata before it introduces you

Every PDF carries data about itself, and it is more revealing than most people expect. There are two layers: the document information dictionary (title, author, creator application, timestamps) and the XMP metadata stream, an embedded XML block that can hold far more — edit history identifiers, the original filename, even remnants inherited from the source document.

PDFs exported from Word are a classic offender. The author field is typically populated with the name from the Office account that created the original document — which may be an employee who left years ago, or the person a "anonymous" report was supposed to protect. Files that went through review cycles can carry traces of comments and tracked changes if they were exported carelessly. Titles auto-filled from the first line of an old template ("Termination Letter — DRAFT — DO NOT SEND") have a way of surfacing in browser tab titles.

Auditing this takes one minute. Open the file and check its document properties (Ctrl+D in Acrobat, or the info panel in most viewers), or run exiftool yourfile.pdf on the command line to dump everything, XMP included. Before publishing anything externally, scrub the fields with your PDF editor's "sanitize" or "remove hidden information" feature. If a document only needs to be read once, viewing it in a private in-browser PDF reader avoids creating yet another metadata-bearing copy in a cloud service's preview system.

4. Watermark for accountability

Encryption controls who can open a file. Watermarking addresses a different problem: what happens after an authorized person opens it. Once a document is legitimately decrypted, nothing technical stops the recipient from forwarding it — but a watermark changes their incentives and gives you a trail if they do.

There are two useful styles, and they serve different goals:

• Visible deterrent watermarks — a diagonal "CONFIDENTIAL — Prepared for J. Smith, Dec 2025" across each page. The point is psychological: a recipient thinking about forwarding a file with their own name stamped on every page usually reconsiders.
• Subtle tracing marks — a small per-recipient code in a footer or margin. When each copy you distribute is uniquely marked, a leaked screenshot tells you exactly whose copy leaked. Law firms and deal rooms have used per-copy marking for decades for precisely this reason.

For board packets, draft agreements, or anything sent to a limited audience, generating per-recipient copies with a watermark tool takes minutes and costs nothing. One caveat: a determined attacker can crop or edit marks out, so treat watermarking as deterrence and attribution, not access control. It complements encryption; it does not replace it.

5. Control the channel, not just the file

A perfectly hardened PDF can still be compromised by how it travels. Email is the default channel and one of the worst: messages are copied onto multiple servers, retained indefinitely, forwarded with one click, and readable by anyone who later compromises either mailbox. An attachment, once sent, is permanently outside your control.

Better options exist and are usually already available to you:

• Expiring share links from storage you control (Google Drive, OneDrive, Dropbox, SharePoint). You can restrict recipients, set an expiry date, disable downloading, and — crucially — revoke access after the fact. An emailed attachment offers none of that.
• Separate channels for file and password. If you must email an encrypted PDF, send the passphrase by text message or a phone call, never in the same thread.
• Be picky about third-party tools. Uploading a sensitive contract to a random free "convert my PDF" site hands a copy to an unknown server with an unknown retention policy — often to save thirty seconds. We covered how these services handle (and mishandle) your files in our breakdown of the hidden dangers of online PDF tools. Prefer tools that process files locally in your browser, so the document never leaves your machine.

Putting it together

None of these steps is exotic, and none requires a security team. Before a sensitive PDF leaves your hands: encrypt it with a real user password and AES-256; remove — don't cover — anything the recipient shouldn't see; check the metadata; watermark copies that go to multiple people; and send it through a channel you can revoke. Five minutes of process prevents the kind of leak that no amount of after-the-fact damage control can fix.

Frequently asked questions

I forgot the password to my own PDF. Can I recover it?

If it is an owner password (the file opens but restricts printing or copying), yes — those restrictions are trivially removable. If it is a user password on an AES-encrypted file, there is no backdoor: your options are remembering the passphrase, finding an unencrypted original, or brute-force guessing, which is only realistic for short, simple passwords. This is by design — an encryption scheme with a recovery loophole would protect nothing. Keep passphrases in a password manager.

Can a PDF contain malware?

Yes. PDFs can embed JavaScript, launch actions, and attached files, all of which have been abused in real attacks — malicious PDFs remain a standard phishing payload. Your defenses: keep your PDF reader updated (most exploits target old, patched vulnerabilities), disable JavaScript in your reader's settings if you don't need it, and treat unexpected attachments with the same suspicion as executable files.

Is a scanned PDF safer to redact, since there's no text layer?

Somewhat, but don't rely on it. A pure image scan has no selectable text under a black box, but many scans have had OCR applied, which adds an invisible text layer that survives your rectangle. And low-quality "redactions" that merely darken pixels can sometimes be reversed by adjusting contrast. Use real redaction or delete the page.

Do I need paid software to do any of this?

Mostly no. Free tools handle page deletion, watermarking, metadata inspection, and private viewing; LibreOffice and macOS Preview can apply user-password encryption at no cost. The one place paid tools earn their keep is high-volume, pattern-based redaction — searching a thousand-page production for every Social Security number is a job for purpose-built software, not a highlighter.