PDF Privacy Cleaner
Deep scan and remove 21 categories of metadata from your PDF — protects your privacy before sharing documents publicly
Drop a PDF here
Select a PDF to scan and remove its metadata
Max file size: 100MB
Remove PDF Metadata Free Online
PDF Privacy Cleaner scans a PDF for hidden metadata and clears the parts you choose, without uploading the file. It reads the document info fields, XMP stream, document IDs, annotations, attachments, embedded photo EXIF and earlier revisions, shows what each category holds, then rewrites the file and rescans it so you can see the result.
Every PDF carries a second layer you never see on the page: the author and title fields your word processor filled in, the name of the software that produced it, document and instance IDs, an XMP stream, EXIF blocks inside embedded photographs, and sometimes annotations, form fields, JavaScript actions or whole attached files. This tool scans for 21 such items, groups them into eight categories, shows what it found, and lets you clear the ones you choose — all inside the browser tab.
The scan runs the moment the file is chosen, so the first thing you see is an inventory rather than a promise: a privacy risk score out of ten, a count of the items found, and a category list you can expand to read the actual values, such as the author name that is really sitting in the file. Cleaning then rewrites the document and rescans the result, and you can download a JSON report of exactly what was removed alongside the cleaned PDF.
How it works
1. Scan the document
Drop a .pdf of up to 100 MB onto “Drop a PDF here”. “Scanning PDF for metadata…” runs, then the Privacy Risk Assessment panel reports the number of items found and a score out of 10.
2. Decide what to clear
Eight categories are listed — Document Info, Document IDs & Revisions, Structured Metadata, Interactive Content, Embedded Content & Fonts, Navigation & Viewing, Output & Print and Incremental Updates — each with a count and an expandable list of what it holds. Everything is selected to start with; Select All and None switch between extremes, and the search box filters long lists.
3. Optionally write new values
With Document Info in the selection, a Custom Replacement Values panel offers Title, Author and Subject fields. Anything you type is written in place of the old value; anything you leave empty is blanked. Creator and Producer are always emptied, never replaced.
4. Clean, verify and download
The action button reads “Clean All Metadata”, or “Clean Selected (3 categories)” when you have narrowed it down. “Cleaning Complete” then reports the size change and how many items were removed, skipped or replaced; “Download Clean PDF” saves the file with -clean in its name and “JSON Report” saves the full record beside it.
When you'd use it
Publishing a document under an organisation's name
A report written on a personal laptop usually carries that account's name as the author. Replacing the Author and Title with the organisation's own values is cleaner than leaving them blank, and takes one field each.
Sending a tender or bid document
Producer strings, document IDs and revision markers can hint at which template a bid came from and how often it was revised. Clearing them leaves the reader with the document and nothing else.
Sharing a PDF that contains photographs
A photo dropped into a document can bring its own EXIF block, including camera model and, on phones, GPS coordinates. The Embedded Content & Fonts category targets exactly that layer without touching the picture itself.
Checking what a PDF is carrying
Even when you clean nothing, the scan is useful on its own: it tells you whether a file you received holds JavaScript, attachments or an editing history before you pass it on.
What the eight categories cover
Document Info is the classic properties panel: title, author, subject, creator, producer and keywords. Document IDs & Revisions covers the document and instance identifiers plus PieceInfo revision markers left by editing software. Structured Metadata handles the XMP stream, the tagged-PDF structure tree, marked-content settings and page labels.
Interactive Content covers annotations, form fields and JavaScript actions. Embedded Content & Fonts covers file attachments, image metadata, font metadata and the EXIF pass over embedded JPEGs. Navigation & Viewing covers bookmarks, named destinations, viewer preferences, optional content groups — layers — and article threads. Output & Print removes output-intent colour profiles, and Incremental Updates drops the earlier revisions that a re-saved PDF can still be carrying.
The EXIF pass is unusual in how it works: rather than re-encoding the pictures, it walks the raw bytes of each embedded JPEG and overwrites the APP1 and APP2 segments with zeros. The image data is left exactly as it was, so picture quality is untouched and the file does not get smaller from that step.
Things that disappear along with the metadata
Two of the categories remove content people often want to keep. Clearing Interactive Content deletes annotations and form fields, so the answers typed into a fillable form go with them — run the form through /pdf/flatten first and the answers become part of the page, after which cleaning is safe. Clearing Navigation & Viewing deletes the bookmark outline and any layers, which matters for long documents people navigate by contents.
Structured Metadata includes the tagged-PDF structure tree. That tree is what lets a screen reader follow headings and reading order, so removing it from a document intended for accessible reading is a real cost. If the document must stay accessible, clear Document Info and the identifiers and leave the structure alone.
Because everything starts selected, the quickest safe habit is to press None and then tick only the categories you actually care about. The counts beside each category tell you whether there is anything in it at all.
What cleaning cannot do
Metadata cleaning never touches what is printed on the page. A name in the letterhead, an address in the footer or a signature block stays exactly where it is, because it is page content rather than metadata. To take visible information out of a document, use /pdf/redact.
The cleaned file is not guaranteed to be smaller — it can come out larger, because the document is re-saved without object-stream compaction. The report shows both sizes, and /pdf/compress is the tool for shrinking the result if that matters.
Verification is honest about its limits too. After cleaning, the tool rescans the output and also searches the raw bytes for strings it saw in the original, listing anything it still finds under a verification entry in the report. Reading that entry is worth the ten seconds it takes when the document is going somewhere public.
An encrypted PDF may fail to scan at all; the panel shows the error rather than a report. Remove the password with /pdf/unlock and clean the unlocked copy.
Frequently asked questions
What is the privacy risk score based on?
It is a summary of the scan: how many metadata items were found across the document and how sensitive their categories are. It is a prompt to look at the list, not a verdict on the document.
Can I keep an author name but change it?
Yes. With Document Info selected, type into the Custom Replacement Values fields for Title, Author or Subject and those values are written into the cleaned file. Fields you leave empty are blanked instead.
Will my filled form still show its answers?
Not if you clear Interactive Content, which removes form fields along with their values. Flatten the form first with /pdf/flatten so the answers become page content, then clean.
Does it remove GPS data from photos inside the PDF?
The EXIF pass zeroes the APP1 and APP2 segments of every embedded JPEG, which is where EXIF and its GPS tags live. The picture itself is not re-encoded, so nothing about image quality changes.
Is there proof the metadata really went?
The tool rescans the cleaned bytes and runs a byte-level search for values from the original scan, then records the outcome in the downloadable JSON report along with every item and action.
Why is my cleaned PDF larger than the original?
Cleaning re-saves the document without object-stream compaction, which can add bytes even as metadata is removed. Both sizes appear in the report; use /pdf/compress if the size matters.
Does this hide who wrote the text on the page?
No. Only the hidden layer is cleaned. Anything visible — a name, a letterhead, a reference number printed on the page — stays, and needs /pdf/redact instead.
Can I clean several PDFs at once?
No, the tool takes one file at a time so the scan, the category counts and the verification report all describe that specific document. Repeat the steps for the next file.