Remove metadata and hidden data from a PDF

See what your PDF is quietly carrying, then strip it out before you send it.

or drop it here · it never leaves your device

Processed in your browser. Your files are never uploaded. Check it yourself

How to remove metadata and hidden data from a PDF

A PDF often says more than its pages do. The document properties can name the person who wrote it, the program they used and the dates it was created and changed. Files can carry attachments, scripts, hidden layers and review comments that never show on the page. Most of it is harmless, but not all of it: a draft that still names its author, a contract with a hidden layer of old notes, a report with someone’s comments left in.

This tool lists what it finds, lets you choose what to remove, and writes a clean copy. It all happens in your browser, so the file you are worried about is not uploaded to do it. If you want to know what kinds of data are lurking, read what hidden data a PDF carries.

1
Open the PDF and read the scan
Press Choose a PDF or drop the file on the panel. The file is scanned on your device straight away, and six rows appear: Metadata, Attached files, JavaScript, Hidden layers, Comments and annotations, and Links. Each row has a count on the right, such as “3 found” or “None found”.

Under any row that found something, Show what was found expands the detail: for metadata, the actual author, title, creating program and dates; for attachments, the file names and sizes; for comments, how many are comments and how many are markup. That detail is displayed on your screen only.
2
Decide what to remove
Metadata, attached files, JavaScript, hidden layers and comments are ticked by default. Links are not, because removing them breaks navigation inside the document and any web addresses people might click. The buttons above the list set the choices in one go: Recommended restores the defaults, Tick all adds links, and Untick all clears every box.

The clean-up button stays disabled until at least one box is ticked, so you cannot download an unchanged copy by accident.
3
Clean the file and read the report
Press Remove hidden data. The file is rewritten from scratch, and a report lists each category with what was found and what was removed, for example “2 of 3 removed” or “None found”. If nothing needed removing, the card says so instead of claiming a result.

The report is the check that matters. A row showing “kept” means that category was found but you left it unticked, and that is worth a second look before the file goes out.
4
Download and verify in another program
Download yourfile-clean.pdf. Open it in a reader and look at File, then Properties (or Document Properties). Author, title and creating program should be empty or generic. If you need to remove visible content too, the result card links straight to Redact PDF.

Remove metadata features

Scans the file and shows what it found before you change anything
Removes author, title, dates, producer and the XMP metadata packet
Removes attached files and embedded JavaScript
Removes hidden layers and the content in them
Removes comments and annotations, and optionally links
A found-and-removed report after cleaning
Writes a fresh file, so earlier saved versions are not carried over

Cleaning a PDF without uploading it

The file is read into this browser tab and processed on your device. Nothing is uploaded, and the details the scan shows you, such as the author’s name, stay on your screen.

The cleaned file is written from scratch instead of being patched, which matters because PDFs can keep earlier saved versions inside them. What you see on each page, the text and the pictures, is not changed.

Metadata covers the document properties (title, author, subject, keywords, creator program, producer, dates) and the XMP packet that repeats them.
Hidden layers are layers switched off by default. Their content is removed, not just switched off again.
Comments are notes, highlights, drawings and text boxes added with a reader. Removing them takes the markup out of the file.
Visible content is out of scope. A name printed on the page stays on the page. To remove that, use Redact PDF.

How to prove your file never left your device: a two-minute Network-tab and offline test you can run on this tool.

Worked example

Sending a draft contract to the other side

Your solicitor’s draft contract is a PDF exported from a word processor. Open it here and the scan reports six document properties (including the author’s full name and the program’s version), one attached spreadsheet, and five review comments. Leave the recommended boxes ticked and press Remove hidden data.

The report shows what was removed: properties and the XMP packet, the attachment, and the comments. Links stay, because they are not ticked. Download contract-clean.pdf and the other side receives the pages and nothing behind them.

Good to know

Run this last. If you add a watermark, compress or merge afterwards, those tools can write new metadata.
Want a proper title and author in the clean file? Set them afterwards with Edit properties.
Keep links unticked for documents people will read on screen. Remove them for files that will be printed or archived.
Metadata is not the only thing that gives a document away. Check the visible text, headers, footers and file name too.
Cleaning a file does not make it confidential. If it must stay private, add a password with Protect PDF.

What is actually stored in a PDF besides the pages

A PDF is a container of numbered objects, and the pages are only some of them. The rest is bookkeeping, and some of it is personal. These are the categories the tool finds and, if you tick them, removes.

Metadata

The document information dictionary holds Title, Author, Subject, Keywords, Creator (the program that made the original), Producer (the program that wrote the PDF) and two dates, created and modified. Word, for instance, takes the author from the name set up in Office, which on a shared machine can be a colleague’s name or a personal account. A second copy lives in the XMP packet, an XML stream that can also hold a document ID and a record of the tools and edits involved. Readers show the dictionary in Properties; some other programs read XMP instead, which is why clearing only one of them is not enough.

Attached files

A PDF can embed other files, listed in an embedded-files table or attached to the page as paperclip annotations. A spreadsheet of working figures or a source Word document can ride along unseen inside a PDF that looks like a one-page letter.

JavaScript

Acrobat-style scripts can run on opening, on printing or when a form field changes. Ordinary documents need none. Scripts are the part of a PDF that security tools care most about, and many mail gateways flag files that contain them.

Hidden layers

Optional content groups, usually called layers, let a PDF hold several versions of a page and switch between them. A layer set to off by default is not drawn, but its content is still in the file. Removing it deletes that content and then drops the layer structure.

Comments and annotations

Sticky notes, highlights, drawn shapes, text boxes and review threads are annotations. They are separate objects floating above the page, so deleting the visible text under a comment does not delete the comment.

Links

Web, email, in-document and launch links are annotations too. They are optional here because they are content as much as data: a table of contents full of links is useful, a link to an internal file server is a leak.

Why the file is rewritten instead of patched

Editing a PDF in place is easy for a program: it appends the changed objects after the old data and writes a new index. Deleting the author name that way changes what a reader shows while leaving the old object in the file. Anyone who knows to look can open it with a hex editor or a tool that walks the file’s revisions.

This tool writes a complete new file containing only the objects the pages still need, so earlier saved versions are not carried across. That is also why the output can be a different size from the input. A file that was saved incrementally many times may shrink noticeably, a file that was already compact will stay close to its original size, and the visible pages, text and images are the same.

The consequence for anyone checking is simple: the cleaned file is a different file, byte for byte, and any digital signature on the original will no longer validate. Keep the signed original, and send the clean copy only where an unsigned copy is acceptable.

When to clean a PDF, and when not to

Cases where it earns its place

Sending a contract or a proposal. The other side does not need to see who drafted it, which version of which program, or the review comments from your legal team.
Publishing on a website. A public file with an internal author name or a draft title is a small privacy leak and an invitation to scrape.
Job applications and CVs. A CV exported from a template may carry the template author’s name in its properties.
Evidence and disclosure. Remove comments and attachments from the copy you serve, and keep a full copy for your own file.
Archiving. Scripts and attachments in long-term records are a maintenance and security risk.

When it is the wrong step

The sensitive thing is on the page. Metadata cleaning never touches visible text or pictures. That is redaction, and the next step is Redact PDF.
The metadata is wanted. A title in the properties is what a screen reader announces and what appears in the browser tab. If your document needs an accessible title, untick Metadata, or set a clean title and author again afterwards with Edit properties.
You need a signed document. Cleaning invalidates signatures.
The file is the evidence. If the metadata itself matters, for example when a dispute turns on who created a file and when, do not alter the original. Work on a copy.

Doing it in Acrobat, Word, Preview or Google Docs

Where	What you can do	Catch
Adobe Acrobat Pro	“Sanitize Document” and “Remove Hidden Information” remove metadata, attachments, scripts, comments and more in one pass.	Paid edition only. The free Reader does not have it.
Microsoft Word	File, Info, Check for Issues, Inspect Document removes personal information from the Word file, and the Save as PDF options let you leave document properties out.	Only works before you export. A PDF you received has no Word file behind it.
Google Docs	Downloading as PDF writes a fresh file with Google’s own producer information, not your editing history.	It still records a creation date and a producer name, and the file title is the document name.
Apple Preview	Shows a document’s properties in its Inspector. Some properties can be viewed there.	It is a viewer for this purpose, not a cleaner: check your version before relying on it to remove anything.
Printing to PDF	Printing a PDF to a new PDF writes a new file with the printer driver’s own properties.	It usually drops bookmarks, form fields and the accessibility tags as well.

A browser’s “Save as PDF” print route does give you a clean-looking file, but it changes more than the metadata, so the document usually loses its tags and bookmarks, and can lose form fields. This tool leaves the pages as they are and removes only the categories you tick.

Two worked examples with numbers

The figures below are round numbers chosen to show the decisions. They are not benchmark results.

A 12-page proposal that went through three authors

The scan reports eight metadata entries (title, author, subject, keywords, creator, producer, created, modified) plus an XMP packet, which the row counts as nine. It reports one attached file, no scripts, one hidden layer in a design export, and 14 comments. Recommended settings remove all of that and leave the proposal’s 31 internal links working. The report reads: Metadata 9 removed, Attached files 1 removed, JavaScript none found, Hidden layers 1 removed, Comments 14 removed, Links 31 kept.

Because the 31 links are shown as “kept”, you know the choice was yours. If the proposal is going to print, tick Links as well.

A form that opens a script

A supplier’s PDF form contains four JavaScript actions that format numbers as you type. You want a plain copy for your records. Untick everything except JavaScript and Metadata. The form fields themselves are not in this tool’s categories and are left as they are; if you also want the answers fixed into the page, flatten the form afterwards.

What this tool does not touch, and edge cases worth knowing

Visible content. The page text, images, headers and footers are never changed here. A person’s name printed in the footer stays in the footer.
The file name. “Smith-v-Jones-version-3.pdf” says plenty. Rename before sending; the download uses your original name plus “-clean”.
Pictures inside the PDF. Some programs embed JPEG photographs byte for byte, which can include the picture’s own camera and location data. This tool does not rewrite image data. If photographs matter, check them before they go into the PDF.
Form values and bookmarks. They are not among the six categories. Bookmark titles can repeat names, so look at the bookmark panel yourself.
Large files. The quick scan that fills in the counts looks at annotations on the first 250 pages, so for a longer file the Comments and Links counts shown before cleaning may be low. The report after cleaning is the one to trust.
Accessibility. Removing the document title can make a file less pleasant for a screen reader user, who hears the title announced when the file opens. Consider adding a plain title back in the source.
Encryption. A password-protected file asks for its password to be opened, and the cleaned copy is saved without one. Add a new password with Protect PDF if the file needs one.

Privacy: scanning a file for personal data should not itself share it

It would be perverse to ask you to upload a document to find out whether it contains personal details. The scan, the display of what was found and the cleaning all run in your browser tab. The author name shown in the “Show what was found” panel is read from your file locally and drawn on your own screen; it is not sent anywhere, not logged, and not kept after you close the tab.

To see for yourself, open the browser’s Network tab, load a file and clean it. You should see the page’s own scripts at first and nothing that carries your document afterwards. The verification guide has the steps, including a test that works with your connection switched off.

Remove metadata: troubleshooting

“Nothing hidden was found in the checked areas”: The file has none of the six categories, or only categories this tool does not check (form values, bookmarks, image data). The file may still be clean. Check its properties in a reader to be sure.
“This file couldn’t be scanned first, but you can still remove everything below”: The quick scan could not read the file, but cleaning can still run. Tick what you want and press the button; the report afterwards shows the true counts.
The report says “kept” against a category: That category was found but its box was unticked. Go back, tick it, and clean again from the original file.
The metadata still shows an author name in another program: Check you opened the clean copy, not the original. Some desktop programs also show the name from a recently opened file’s cached information, so close and reopen the file.
A digital signature now shows as invalid: Cleaning rewrites the file, which breaks any signature on it. Keep the signed original for the record and send the clean copy only where an unsigned file is acceptable.
The links in my document stopped working: Links were ticked. They are off by default. Clean the original again with Recommended settings.
“This PDF needs a password”: Enter the password at the prompt. If you do not have it, nothing here can open the file; Unlock PDF only removes passwords you already know.
“This PDF looks damaged”: Run Repair PDF first, then clean the repaired copy.
The file got bigger after cleaning: Rewriting can store the same data less tightly than the original, which was sometimes optimised. Pass the result through Compress PDF if size matters.

Glossary

Metadata: Data about the document rather than its pages: title, author, program, dates and keywords.
Document information dictionary: The older place a PDF stores its title, author and dates. Readers show it as Properties.
XMP: Extensible Metadata Platform: an XML stream, often duplicating the information dictionary and sometimes adding edit history and document IDs.
Producer and Creator: Creator names the program that made the source document. Producer names the program that wrote the PDF file.
Optional content group: The PDF name for a layer. Content in a group switched off by default is still in the file.
Annotation: An object laid over a page: a note, highlight, drawing, link or form widget. Not part of the page text.
Incremental update: A save that appends changes after the old data. Old revisions can survive inside the file.
Embedded file: Another file stored inside a PDF, listed in the file’s attachments.

Remove PDF metadata: frequently asked questions

What does this tool remove?

The document properties (title, author, subject, keywords, creator, producer and dates), the XMP packet, attached files, JavaScript, hidden layers, comments and annotations, and optionally links. You choose which.

Will it change how my PDF looks?

No. Pages, text, images and fonts are left alone. The only visible difference is if you remove comments, links or hidden layers that you could see in a reader.

Can I see what is in the file before I remove it?

Yes. The file is scanned when you open it and each kind of data shows how many were found, with details you can expand. Those details stay on your screen.

Does it remove text I can see, like a name on the page?

No. That is page content, not hidden data. Use Redact PDF to remove visible text and images permanently.

Why remove JavaScript from a PDF?

Scripts can run when a file is opened, and some readers handle them badly. Almost no ordinary document needs one, so removing it makes the file safer to send.

Is my file uploaded?

No. The scan and the cleaning both run in your browser. Nothing leaves your device, which you can check yourself.

Does it remove older versions saved inside the file?

The cleaned PDF is written afresh instead of having changes added on the end, so earlier saved revisions are not carried across.

Will it work on a password-protected PDF?

You will be asked for the password to open it, and the cleaned copy is saved without one. Add a new password afterwards with Protect PDF if you need it.

Why are links left alone by default?

Because removing them stops people clicking through to web pages and jumping between sections. Tick Links when the file will be printed or archived and the links have no use.

Does cleaning break a digital signature?

Yes. The file is rewritten, so a signature on the original no longer validates on the copy. Send the signed original wherever a valid signature is needed.

Can it remove the author from a Word document?

It cleans PDFs. For a Word file, use Word’s Inspect Document feature before you export, or clean the PDF here afterwards.

Is a cleaned PDF anonymous?

Not necessarily. The visible text, the file name, embedded photographs and the writing style can all identify an author. This tool removes the hidden data it lists and nothing more.

Page updated 28 October 2026 · By the PDFMagpie team