How to Remove Metadata from a PDF

What “PDF metadata” usually means

People say metadata when they mean the extra facts stored with a file: who wrote it, which program produced it, when it was saved, and sometimes keywords or a subject line. Adobe’s public hub article on removing PDF metadata describes those fields in that way: author, title, creation date, keywords, and the software used to make the file.

Inside a PDF those facts are not one bucket. The common properties dialog reads a Document Info dictionary. A separate XML packet called XMP can hang off the document catalog, off individual pages, or off other objects. Clearing the fields you can see in a properties window does not automatically clear every XMP packet. If a guide skips that split, it is describing a dialog, not the file. The PDF metadata vs XMP comparison walks through that coexistence in more detail.

This site’s viewer reports standard Document Info values, the number of Info entries, whether document-level XMP is present, how many pages carry their own XMP stream, page count, and whether a signature field exists. It does not parse every XMP property or every custom object. That report is the honest starting point before you remove anything.

PDF metadata privacy: what can leak when you share

Visible pages can look ready for a client, a public download, or a hiring manager while the file still carries an internal author, a draft title, a producer string from last year’s template, or dates that reveal a workflow. Metadata2Go’s explainer lists author names, software, and hidden document history among the PDF examples people try to strip before publishing.

Privacy risk is not limited to those Info fields. A vendor-authored 14 August 2026 comparison published by Lyonite hid ordinary document properties, a geotagged photo, a named review comment, an attachment, and a script in one PDF, then ran popular “remove metadata” tools. The page also promotes Lyonite’s own remover, so treat its results as vendor evidence. In that comparison, most online removers cleared or attempted the document properties layer and left the other payloads in the file. The lesson is not that every PDF contains GPS. The lesson is that the phrase “remove metadata” is easy to over-read.

Decide the audience first. A resume sent outside your company may only need Title, Author, and detectable XMP gone. A regulated handoff that must not carry attachments, scripts, or embedded-image EXIF is a Sanitize-class job. This tutorial keeps those jobs apart so you can pick a method that matches the actual risk.

Check PDF metadata before sharing

Checking is cheaper than guessing. Open the exact copy you plan to send, not an earlier export sitting in another folder. Look at Title, Author, Subject, Keywords, Creator, Producer, and the two dates. The PDF metadata fields reference lists those eight keys, their usual sources, and whether this site can edit them. Note whether document or page XMP is present. Note whether a signature field exists, because a later rewrite can break that signature.

Use the PDF metadata viewer for that read-only pass. The viewer does not write a new file. If the report is already empty in the supported structures, you may not need a remover. If a signature field is present, stop and follow the signing authority’s process instead of forcing a cleanup.

Checking also catches the opposite problem: a file that still needs a title or author for an archive, catalog, or accessibility workflow. Removal is not automatically the private choice. Sometimes the next owner needs those fields, and the private choice is to edit them on the PDF metadata editor while leaving XMP alone.

Method 1: remove supported metadata in the browser

This site’s remover is the method we can describe from running code, not from a vendor screenshot. Choose one PDF up to 25 MB. The file is read in browser memory. There is no PDF upload endpoint and no account. The tool deletes the complete Document Info dictionary, removes detectable XMP references attached to the document catalog or individual pages, saves a new copy, and reparses that copy. Download is offered only when Info entries are gone and those supported XMP references are no longer detected.

1. Inspect the file in the viewer if you have not already.
2. Open Remove metadata from PDF and choose the same local file.
3. Review the detected Info count and XMP flags, then run Remove and verify metadata.
4. Download only after the page reports that the saved copy passed the supported check. Keep the original until you have opened the download.

The verified scope stops there. Attachments, annotations, visible text, OCR layers, embedded-image EXIF or GPS, scripts, custom objects, every earlier revision artifact, and copies already stored by another service are outside the check. Encrypted PDFs fail immediately. Signed PDFs can be inspected and will not be rewritten here. That is a privacy feature of the write boundary, not a claim that the signed file is clean.

Method 2: desktop properties versus Sanitize

Adobe’s hub walkthrough for “how to remove metadata from a PDF” uses Document Properties: open the file, choose Menu or File, open Document Properties, edit or delete the listed fields, confirm, and save. The same article says this will delete all the metadata in the PDF. Treat that sentence as Adobe describing the properties dialog, not as a lab result about every hidden object.

Adobe Help documents a different product action: Sanitize PDFs in Acrobat Pro. That page describes removing hidden content, including metadata, embedded content and attachments, scripts, and hidden layers. Lyonite’s vendor-authored comparison explicitly did not run Acrobat Pro, so this tutorial will not invent a score for Sanitize. The documented scope is still wider than Document Properties and wider than this site’s remover.

Use the properties dialog when you own the file, can save it, and only need the visible descriptive fields changed. Use Sanitize, or another specialist workflow your organization validates, when attachments, scripts, or other hidden content are in scope. Use this site’s local remover when the job is Document Info plus detectable document/page XMP and the file can stay on your device.

Does removing PDF metadata remove XMP?

Not by default. Document Info and XMP are separate storage. Clearing Title and Author can leave an XMP packet that still names a person, a tool, or a date. This site’s editor is explicit about that: it updates standard Info fields and preserves existing XMP. The remover is the tool that also targets detectable document-level and page-level XMP references and then checks that those references are gone.

“Detectable” matters. The code looks for a Metadata stream on the catalog and on page dictionaries. It does not promise to find every custom object that happens to contain XML. If two structures share the same XMP object, the remover unlinks the catalog and page references it knows about; it does not claim to erase every other pointer in the file.

If XMP is the reason you opened this article, inspect first. When the viewer says document XMP is detected or lists pages with XMP, a properties-only cleanup is the wrong method. Run the scoped remover, then inspect the download. If XMP is still required for a publishing pipeline, do not remove it.

Risks that keep showing up in cleanup advice

Print to PDF is a frequent shortcut. Adobe’s hub FAQ says you can print to PDF to save a flattened copy without metadata. Printing creates a new file, so some source fields can disappear, but the new file can also pick up a producer from the driver, lose forms or links, or rasterize content. Use it only when you will inspect the result the same way you would inspect any other export.

Command-line folklore has the same problem. Lyonite’s retest of common advice found that `exiftool -all=` writes an incremental PDF update: the old values can still be recovered, and ExifTool itself warns that those edits are reversible. Linearizing with qpdf afterwards made the Info-layer change stick in that test, and still left embedded GPS, comments, attachments, and scripts. Ghostscript’s usual pdfwrite invocation carried the Info dictionary across unless extra pdfmark work was added. Quote those tools only with their actual behavior.

Upload-based removers add a disclosure step that this site refuses for PDFs: the file leaves your machine. Metadata2Go’s own blog says complete removal is not always possible because of file structure. That honesty is useful. It is also a reminder to read the scope before you send a sensitive document to a server that promises a clean file.

Two local limits stay in force on this site. Encrypted PDFs are rejected because the browser tool has no decryption workflow. Signature fields stop writing because any rewrite can invalidate the signature. Those are fail-fast boundaries, not incomplete features waiting for a hidden override.