Remove the Metadata That Names You Before Double-Blind Review
You took your names off the title page. The PDF's properties and the photos in your figures may not have — and those are the places reviewers can check in seconds.
Anonymising the text is not anonymising the file
Double-blind venues ask you to remove author names, affiliations and acknowledgments from the paper. That handles what a reviewer reads. It does not handle what the file records about itself, and a reviewer who opens the document properties sees that in two clicks.
The risk is not hypothetical. In one study of 43 anonymous PDF submissions to a conference, three were not properly anonymised: one carried the authors' names in a metadata field, and two leaked identity through metadata in the images embedded in the paper.
Where your name hides in a PDF
The document's own metadata sits in four places, and clearing one rarely clears the others:
- The Author field — filled from your user name by Word, and by LaTeX when a template or hyperref's pdfauthor and pdfusetitle options set it from \author.
- Creator and Producer strings, which name the software and can include file paths or a shared drive's host name.
- An XMP metadata packet, which repeats the document properties and can hold more.
- Old revisions. A PDF edited and re-saved can keep earlier versions inside the file, with the author fields as they were before you cleared them.
Your figures carry their own metadata
PDF builders, LaTeX included, commonly embed JPEG figures without re-encoding them, which means each photo's metadata travels into the PDF with it: the camera or phone that took it, the time and place, and in images edited in Photoshop or Lightroom, the creator name and editing history in XMP.
Tools that clean a PDF's properties usually leave those embedded images alone — the PDF cleaner on this site included, which copies images through untouched so the pages are not altered. So the figures have to be cleaned before they go into the paper, not after.
A workflow that covers both
Do it in this order, because each step writes new metadata that the next one removes:
- Clean every figure image before you build: JPG, PNG, WebP and HEIC files go through the image cleaner, which removes EXIF, XMP and IPTC without re-compressing them. Figures regenerated from your plotting code start clean anyway.
- Build the anonymous version with the anonymity your template provides — the class's review or anonymous option in LaTeX, and no pdfauthor. In Word, use Inspect Document to remove document properties and personal information before exporting.
- Clean the final PDF: the PDF cleaner clears the document information, the XMP packet and old revisions in one pass, then re-reads the file to confirm.
- Open the result's document properties in your viewer and check them yourself before uploading.
What metadata cleaning does not cover
Identity also leaks through the content: self-citations written in the first person, acknowledgments, grant numbers, URLs with your username, and supplementary code or data with names in licence files. CVPR's guidelines, for example, ask authors to remove names and institutions from submitted code and data, copyright and licence statements included.
Check the current call for papers for your venue's exact rules. Metadata is the part a checklist most often misses, not the only part.
Clean your submission PDF
PDFs record who made them, with what software and when — and often keep earlier drafts inside the same file. See what yours contains, then remove it.
Open Remove Metadata From a PDFFrequently asked questions
Does LaTeX put my name in the PDF properties?
It can. hyperref writes an Author field when pdfauthor is set, or from \author when pdfusetitle is on, and some templates do this for you. Use your class's anonymous option and check the properties of the built PDF.
Do images inside a PDF keep their EXIF data?
Often, yes. JPEG figures are commonly embedded without re-encoding, metadata included, and PDF cleaners usually copy images through untouched. Clean the figure files before building the PDF.
How do I remove my name from a Word document's properties?
Use Inspect Document in Word's file menu and remove document properties and personal information, then export the PDF. Check the PDF's properties afterwards, because export settings can add their own.
Can reviewers see old versions of my PDF?
Sometimes. A PDF saved incrementally keeps earlier revisions inside the file, so cleared author fields can survive in an older one. A cleaner that rewrites the file removes them.
Does removing metadata make my paper anonymous?
No, only the file. Self-citations, acknowledgments, usernames in links and names in supplementary material identify you just as well, and need the venue's own anonymisation rules.
Should I clean supplementary files too?
Yes. Images, PDFs and archives in supplementary material carry metadata in the same way, and venues such as CVPR ask for names to be removed from code and data, licences included.