How to Remove Metadata from a PDF (Without Uploading It)

Remove PDF metadata in three steps

1. Open the scrubber. Zero-Upload PDF Metadata Scrubber — no install, no account, no upload.
2. Drop the PDF in (or click to browse). The page parses the file locally and prints what it is deleting, line by line, into a terminal log.
3. Download the cleaned file. The tool re-parses its own output before offering it, and prints the proof: verify: /Info keys remaining = 0 · XMP payload = empty.

What it clears: Title, Author, Subject, Creator, Producer, CreationDate, ModDate, Keywords and any custom keys a workflow bolted on. What it leaves alone: pages, text, images, links, form field values, bookmarks and attachments — the same document, minus the paper trail.

Two caveats worth knowing before you start. An encrypted PDF is rejected, because the tool does not re-encrypt files. And because the scrubber removes everything, a PDF you intend to publish loses the Title that search engines use as its headline — if the file will be indexed, set a clean Title afterwards in your editor of choice.

What a PDF hides: the field-by-field list

Metadata lives in the document's information dictionary, one entry per line. This is what each one exposes about you:

Then there is the second copy. Every modern PDF also carries an XMP packet — the same information again, in XML, referenced from the document catalog. It routinely holds more than the dictionary does: edit history, software build numbers, colour profiles, and — in files that came off a network scanner or a multifunction printer — the device name and the operator's login. Clearing the visible Description fields and leaving XMP intact is the single most common reason “I already removed the metadata” turns out to be wrong.

Check a PDF's metadata in 30 seconds

Before you send anything, open it somewhere and read the record:

• Acrobat or Acrobat Reader: File → Properties → Description. The XML copy sits under File → Properties → Additional Metadata → Advanced.
• Chrome or Edge: open the PDF in the browser and use Document properties in the viewer sidebar — Title, Author and both dates are shown.
• Windows Explorer: right-click the file → Properties → Details.
• macOS Preview: Tools → Show Inspector (⌘⌥I).
• Terminal: pdfinfo file.pdf for the summary, or exiftool file.pdf for a full dump including XMP.

If any of those shows a name, a date or a software version you did not intend to publish, the file is not ready to go out.

Manual removal — and where each method stops

• Acrobat Pro → Protect → Sanitize Document. The most complete built-in pass: metadata plus other hidden data, in one action. The free Reader can view metadata but cannot remove any.
• Acrobat Pro → File → Properties. Clear the Description fields by hand, then open Additional Metadata → Advanced and delete the XML. Two separate places — which is exactly how people end up cleaning one of them.
• Windows Explorer. Right-click → Properties → Details → Remove Properties and Personal Information → “Create a copy with all possible properties removed”. It only touches what Windows understands; Producer and the XMP packet usually survive.
• Preview (macOS). Reads most of it, clears very little: Producer, Creator and the XMP packet stay put.
• Re-export from Word or Google Docs. Genuinely effective, because the PDF is rewritten from scratch — but the layout is re-set (fonts, vector detail, form fields), so it is a bad trade for a contract or a print-ready file.
• Command line: exiftool -all= -overwrite_original file.pdf strips the dictionary and XMP together. Leave off -overwrite_original first if you want a safety copy alongside it.

Every one of these shares the same failure mode: there are two copies of the metadata, and any method that only clears the visible fields leaves the second one behind. Verify after every method, not just the automatic ones.

Why the metadata comes back after the next save

Because the next application to write the file stamps it again. Save a scrubbed PDF from Word and /Producer becomes “Microsoft® Word for Microsoft 365” with a fresh /ModDate. Export it from Pages, LibreOffice, LaTeX or a print-to-PDF driver and it re-stamps its own Creator and Producer — sometimes with your macOS user name in the Author slot, read straight from the machine.

So position the scrub correctly in the workflow: final draft → export → scrub → send. Not scrub → open → edit → save. A scrub is permanent across sends, not across edits — and any tool that promises otherwise is describing a one-time state, not a property of the file.

What metadata removal does not do

Being precise about the limits keeps this from becoming false comfort:

• It is not redaction. Page content stays selectable and searchable. A name typed on page 3 is still findable — redact the page itself.
• Attachments, bookmarks, form field values, comments and hidden layers are untouched. Document-level metadata is a different object from document content.
• It breaks a digital signature. Any byte change invalidates a signature — leave signed filings alone, or re-sign after scrubbing.
• Image-level EXIF inside embedded JPEGs is not touched. That is a separate record, inside a separate stream.
• Metadata is not identity. File name, fonts, wording, watermarks and revision history inside the content can still point straight at you.

Verify it worked, then send

Run the same check you did in the beginning — properties dialog, pdfinfo, or exiftool. A clean file shows almost nothing: page count, page size, file size, PDF version. The Description tab is empty, the advanced XML window is empty, and no Author line is rendered anywhere. The scrubber prints the same verdict from inside the log before it offers the download, so you never have to take the result on trust.