Erasemetadata

Remove metadata from PDF files

A PDF usually names the person who wrote it, the application that produced it — with a version — and often the original Word or InDesign document it was exported from. It can also carry an XMP block with a full editing trail. This tool empties both without touching page content, and it refuses outright to edit a PDF that is digitally signed, because that would invalidate the signature.

  • No upload
  • No account
  • No re-encoding
  • Free
Input0 bytes uploaded

Engine

File API

Runs in

Web Worker

Needs account

No

What a PDF usually carries

01

Two places hold this data: the document information dictionary (the classic fields below) and an XMP metadata stream, which is an XML document that can hold far more.

  • AuthorThe name or account that created the documentmedium
  • Title, subject, keywordsFree text, often a client name or a project codelow
  • Creator applicationThe app that produced the source document, e.g. Word or InDesignlow
  • ProducerThe library or export tool, with a version numberlow
  • Creation and modification datesLocal timestamps recorded at exportlow
  • XMP metadata streamAn XML block: edit history, tool chain, sometimes AI labelsmedium
  • Piece informationPer-page data some editors attach for round-trippinglow
  • Digital signatureNot a privacy field — a reason this tool will not modify the filekept

Why it matters.

02

Contracts and invoices name the author

The author field is whatever the exporting application wrote: an employee name, a client name, a template author. Invoices sent to customers routinely carry the internal account that generated them.

The producer line dates the file and names the toolchain

“Adobe PDF Library 15.0”, “Microsoft Word 2016”, “Skia/PDF m121” — each tells a reader which app and roughly which era the file came from, and sometimes which operating system. Combined with timestamps it narrows down a specific machine.

XMP is where the history lives

Many editors write an XMP packet alongside the classic fields: the source document, the export settings, the app that saved it last, occasionally the AI tool that generated an image placed inside. Emptying the information dictionary alone leaves that stream behind.

A signed PDF is a different problem

A signature covers the file as it was at signing time. Removing metadata changes the bytes, so the signature stops validating — and a contract that opens with “signature invalid” is worse than one with an author name in it. This tool refuses instead of quietly breaking it.

How to use it.

03
  1. Step 01

    Drop a PDF into the panel. It is parsed in a Web Worker on your device.

  2. Step 02

    Check the report. A signed file is flagged here and cannot be cleaned.

  3. Step 03

    Download the cleaned copy and open it once to confirm the pages still render as expected.

One caveat.

04

Before you rely on it

Rebuild, do not patch

A PDF is a graph of objects joined by a cross-reference table. Deleting a metadata object in place means rewriting offsets and risking a file that opens in one viewer and fails in another. This tool parses the document, empties the two metadata carriers and writes a new file, so the result is a valid PDF that any viewer accepts — at the cost of a new byte layout.

Limits worth knowing.

06
  • Digitally signed PDFs are refused. Flatten or re-export the document first if you still need its metadata gone.
  • Encrypted PDFs cannot be rewritten without the password.
  • Form field values and annotations are part of the document, not of its metadata, and are left alone.
  • A PDF is re-serialised rather than patched, so the file gets a new byte layout. Text, fonts and images are copied unchanged, but the file size can shift either way.
  • Cameras and scanners burn EXIF-like data into images *inside* a PDF. Those pixels are content and are not touched.

Questions people ask.

07
Will the PDF still open after cleaning?

It should. Only the information dictionary and the XMP stream are emptied; pages, fonts, images and the outline are copied as they were. As with any file editor, keep the original until you have opened the cleaned copy once.

Why was my signed PDF refused?

Because a signature is a hash of the whole file. Any change — including removing an author name — invalidates it. Silently producing a file whose signature no longer verifies would be worse than saying no.

Does this remove the text from the PDF?

No. Page content is untouched, including scanned images and their own embedded EXIF. This tool only handles the document-level metadata.

The author field came back after I edited the PDF again

Most editors rewrite document properties when they save. Clean the file last, or clean it again after the final export — the same is true for Office documents.

Is the PDF uploaded?

No. Parsing and rebuilding happen in a Web Worker in your browser. There is no upload endpoint; the network panel will show no request carrying the file.

Related tools.

08
Files are processed on your device. Erasemetadata cannot read them.PDF