BASALT · JOURNAL
Your PDF is telling everyone who wrote it, and where
When a PDF shows author name in a reader's properties panel, that string comes from the /Author entry in the document information dictionary, and usually from a duplicate inside the XMP metadata packet as well. It is not on any page, no amount of visual review will surface it, and it survives almost every redaction workflow.
The name is rarely the only thing in there. The same two structures typically carry the producing application, the original file path, prior titles, creation and modification timestamps, and an identifier linking this file to every earlier version of itself.
Why your PDF shows author name
Two separate metadata systems live in a PDF and both are populated automatically.
The first is the document information dictionary, an object referenced from the trailer with keys such as /Title, /Author, /Subject, /Creator, /Producer, /CreationDate, and /ModDate. /Creator names the application that produced the original content and /Producer the one that wrote the PDF bytes. /Author is filled from whatever identity the authoring application knows about you, which on a Mac is often your full name from the system account or from the Word or Pages user information.
The second is XMP, an XML packet embedded as a stream in the document catalog. It carries the same fields under different names, dc:creator for the author and xmp:CreatorTool for the application, plus a history array that some applications fill with a record of edits and the software that made them.
The two systems are independent, which is why they disagree so often. A tool that clears /Author in the info dictionary and never touches the XMP packet produces a file reporting no author in one panel and your full name in another.
A third field is worth knowing about. The trailer's /ID array holds two identifiers, the first fixed at creation and the second updated on each save. The first persists across versions, so two documents you believed unrelated can be tied together by a matching identifier long after every visible field is cleared.
Where else the name is hiding
Clearing the two metadata systems is necessary and not sufficient, because names get written into other structures that no properties panel displays.
Annotation dictionaries carry a /T entry holding the title of the annotation, which authoring tools fill with the commenter's name. A document that went through review can carry the name of every person who left a sticky note, in plain text.
Signature dictionaries record the signer's name, contact information, and often the reason and location fields. Form fields keep their values in /V entries in the field dictionary rather than page content, so a name typed into a form persists after the appearance is covered.
Embedded file attachments carry their own metadata. A spreadsheet attached to a PDF brings its own author, company, and edit history with it, and pdfdetach -list is the only way you will see it.
Then there are the structures generated during editing: stale page thumbnails rendered before your last change, optional content group names describing layers you switched off, and JavaScript actions that reference paths or user names. What a PDF still carries after you think you cleaned it inventories the full set.
Confirming exactly what is in your file
Install exiftool, qpdf, and Poppler with Homebrew. The first command reads both metadata systems and groups them so you can see where each value comes from.
exiftool -a -u -g1 document.pdfValues under the PDF group come from the information dictionary and values under XMP groups from the packet, so a name in only one of them tells you a previous cleanup was partial.
To see the raw structures rather than a summary, expand the file and read it.
qpdf --qdf --object-streams=disable document.pdf expanded.pdf
strings expanded.pdf | grep -i "your name"That search catches names in annotation /T entries, signature dictionaries, form field values, and optional content group names, which summary tools do not consistently surface. Look for the trailer /ID array and the catalog's /Metadata while you are there.
List attachments separately, since they are invisible in normal viewing.
pdfdetach -list document.pdfCount generations, since a cleaned current revision does not help if an earlier one is still in the file.
strings document.pdf | grep -c "%%EOF"More than one end-of-file marker means incremental updates are present and a previous revision may hold the metadata you just cleared. Inspecting what is inside a PDF covers reading the expanded structure.
Stripping it so it stays stripped
Clearing metadata correctly means treating both systems as one operation and writing the result as a fresh generation.
The information dictionary entries have to be removed rather than blanked, the XMP packet removed or rewritten rather than emptied, and the two done together so they cannot disagree. Annotation /T entries, form field values, and signature dictionaries need handling explicitly, since none of them are metadata in the sense a metadata tool means.
The file then has to be written as a single generation with a new cross reference table. Appending the cleanup as an incremental update leaves the old information dictionary and XMP packet physically present with an earlier cross reference table pointing at them, so the cleanup is undone by anyone who walks the update chain back one revision.
The trailer /ID array should be regenerated rather than carried forward, so the output cannot be correlated with earlier versions. Stale thumbnails should be dropped rather than rebuilt from a cached copy.
Then re-run the same commands on the result. Verifying with the tool that performed the cleanup only tells you what that tool believes it did. Removing PDF metadata on a Mac covers the mechanics field by field.
Frequently asked questions
Why does my PDF show an author name I never typed?
Because the authoring application filled the /Author entry automatically from your system account name, your Office or Pages user information, or the login of whoever created the original template. PDF generators propagate these values without asking, and the same name is usually duplicated into the XMP packet as dc:creator.
How do I remove the author name from a PDF on a Mac?
Use a tool that clears both the document information dictionary and the XMP packet in the same pass and writes the output as a single generation. Removing the entries rather than blanking them matters, and so does regenerating the trailer /ID array. Verify afterward with exiftool -a -u -g1 and by expanding the file with qpdf.
Does saving a PDF from Preview remove the metadata?
No, in general it preserves it. Re-saving from a viewer carries the information dictionary and XMP packet forward and may add a /Producer entry naming the viewer. It can also write the result as an incremental update, which leaves the original metadata objects in the file. Assume nothing was removed unless you have checked.
Where does the PDF creator field come from?
/Creator names the application that produced the original content, such as a word processor, and /Producer names the software that wrote the PDF bytes, often a separate export engine or print driver. Together they describe your toolchain, and they frequently reveal the version you were running. XMP stores an equivalent value in xmp:CreatorTool.
Can metadata reveal my file path or computer name?
Yes, and it often does. XMP history arrays and some producer fields record the full path of the source document, which can include your account name, a client name in a folder, or an internal project code. Attached files carry their own paths. Read the complete exiftool -a -u -g1 output rather than the summary a viewer shows.
Does redacting a PDF remove the author name?
Not unless the redaction tool explicitly strips metadata in the same operation, and many do not. Redaction usually targets page content, while the information dictionary and XMP packet sit outside the page tree entirely. A perfectly redacted document can still name its author, its firm, and the path it was drafted in.
Why does exiftool show two different authors for the same file?
Because the document information dictionary and the XMP packet are separate structures that can hold different values, and a partial cleanup or a chain of editing tools leaves them out of sync. Neither is authoritative, and different readers display different ones. Treat any discrepancy as evidence that a previous cleanup touched only one system.
Will stripping metadata break a signed PDF?
Yes, it will invalidate any existing digital signature, because the signature covers a byte range of the file and rewriting the document changes those bytes. Strip metadata before signing, not after. If a signed document needs cleaning, produce a clean version from the source and sign that instead.
Doing it in Basalt
Basalt strips the document information dictionary and the XMP packet together, along with embedded attachments, hidden layers, annotation values, and stale thumbnails, and writes the result as a single generation so the cleared values cannot be recovered from an earlier revision. The Inspector shows what a document carries off-page before you change anything: metadata, XMP, embedded files, hidden layers, annotations, OCR layers, JavaScript, and saved generations. Every tool writes a new file and the original is never modified, so you can compare the two. The engine has no network entitlement, enforced by macOS at the code signature level, so documents never leave your Mac, and there is no account and no telemetry. It is $29 once, lifetime, up to three Macs, free for 24 hours.
Redaction that proves itself
Basalt destroys the content you mark, then re-opens the file it wrote and proves the content is gone before it saves anything. Your documents never leave your Mac.