Anonymise a PDF for blind review or screening, every trace in order

Remove metadataAuthor, dates, XMP, hidden data. Free, runs on your device.Open

The 5 places identity leaks

Place	What gives you away	The fix	The check
1. The text	Your name, employer, city, grant number, “as we showed in our 2024 paper”	Rewrite in the source; redact what you cannot change	Search the text for your name and employer
2. Headers, footers, title page	A running header with your name, a logo, a page footer with a department	Remove in the source; redact on every page	Look at the first, a middle and the last page
3. Metadata	Author, Creator (often your institution’s licensed software), XMP	Remove PDF metadata	The result card, then Inspect
4. Outside the pages	Comment authors, bookmark titles, attached files	Delete in the source, or a fresh copy with Extract pages	Inspect: Annotations, Outline, Attachments
5. The wrapper	The file name, the email or portal account it comes from	Rename; submit through the portal’s anonymous route	Read the file name aloud

Pass 1: the text

The cleanest anonymous PDF is one that never had the name in it. Open the source (Word, LaTeX, Google Docs), and:

• Replace your name and co-authors’ names with the venue’s placeholder (“Author 1”, “Anonymous”).
• Cite your own earlier work in the third person (“Cooper (2024) showed”), or as the venue asks (“Anonymous, 2024”).
• Remove acknowledgements, funding numbers and ethics approval numbers, or move them to a separate file the venue keeps back.
• In Word, accept or reject all tracked changes and delete comments before export. A comment carries its author’s name into the PDF.
• Export to PDF again.

Where you cannot touch the source, for example a scanned letter of support or a signed form, redact:

1. Open the PDF at /redact-pdf. Nothing is uploaded.
2. Drag a box over each name, signature, logo and address. Find personal data can mark emails and phone numbers for you; it does not find names, so those are by hand.
3. Click Save. The text under each box leaves the file, character by character, and the rest of the line stays text. The black bars announce that something was removed, which is normal in screening and fine for most reviews.

Pass 2: the metadata

Drop the PDF on Remove PDF metadata. Leave Title set to “removed too” unless the title says nothing about you, and click Remove metadata. It removes:

• every document property: Title, Author, Subject, Keywords, Creator, Producer, both dates, and any custom entries an application added;
• the XMP packet, the second copy of these facts that a properties dialog can miss, including per-page packets;
• PieceInfo, where some applications keep private data;
• the file identifier, which it replaces, because it can be derived from the original’s dates;
• leftovers nothing points at any more, such as earlier versions of pages edited with an appending save, which can still hold a name you deleted. The card says how many it removed.

The result card lists what it found before it removed it, and says “Removed 1 XMP metadata packet” when there was one. Our test file had the author in both stores; the cleaned copy had neither, and no trace of the name in its bytes. The output is saved as -clean.

Pass 3: outside the pages

Three things a reviewer’s viewer can show that are not on any page:

• Comments and notes. Each one stores its author’s name. Delete them in the source before export, or open the PDF in the editor, select each note or highlight and delete it. Inspect counts them in the Security panel under Annotations.
• Bookmarks. A bookmark titled “Chapter 3 (draft for Prof. Čermák)” is visible in every viewer’s sidebar. Inspect lists them under Outline.
• Attachments. A supplementary spreadsheet embedded in the PDF keeps its own author and content. Inspect lists them under Attachments.

Remove PDF metadata leaves bookmarks and attachments alone (we checked: both survived). To drop both at once, run Extract pages with Pages set to “1-”: it writes a new file with every page and, in our test, no bookmarks, no attachments, and only “getPDF” as the producer in its properties.

Pass 4: the wrapper

Rename the file before you upload it: “submission-1234.pdf” or whatever the venue asks for. Portals often show the file name to reviewers, and “cooper-graz-lab-v7-final.pdf” undoes everything above. If the portal keeps your account details separate from what reviewers see, use its upload route, not an email to the editor.

Check it, the way a curious reviewer would

1. Inspect it. Drop the final file on Inspect. Metadata and XMP metadata should say none; Outline should say none or list neutral titles; Annotations and Attachments should be 0 or neutral.
2. Search the text. Run PDF to Text on it and search for your surname, your co-authors’, your institution, your city and your grant number.
3. Ask a colleague outside your group to read the first 2 pages cold and guess who wrote it. Five minutes, and the most honest test there is.

The honest part

Tools remove data, not fingerprints. Your topic, your method, your writing style, the data set only your lab has, and a reference list heavy with 1 group’s papers can identify you to a reviewer in your field, and no PDF tool changes that. Double-blind review reduces bias; it does not guarantee anonymity, and venues know it.

Two smaller limits. Redaction is visible: a reviewer sees black bars and can guess what length of name sat there, so rewriting beats redacting. And Remove PDF metadata does not look inside pictures: a photo or figure embedded with its own camera or software metadata keeps it, so export figures from your plotting tool without author details, or paste them as fresh images. What PDF metadata reveals covers that layer in detail.