Remove PDF Metadata

Every PDF carries a set of properties describing the document rather than its content, and they are filled in automatically by whatever produced the file. The author field takes the name from the account that created it. The creator and producer fields record the exact application and version. Timestamps record when it was made and last changed, sometimes to the second.

None of that is visible on the page, and almost nobody checks it before sharing. It is, however, trivially readable — a right-click in any PDF viewer, or a search engine indexing the file. Documents have revealed the real name behind an anonymous submission, the identity of a whistleblower, and the fact that a bespoke proposal was produced from a template with another client's name still in the author field.

The template case is the most common and the most quietly embarrassing. A tender response, a contract or a report generated from a colleague's file inherits their name, which means your document is telling the recipient who really wrote it.

PDFs store metadata twice, for historical reasons. The document information dictionary is the original mechanism: a simple list of title, author, subject, keywords, creator, producer and dates. The XMP packet is the newer approach, an embedded XML block that usually repeats those values and can carry considerably more, including rights statements and tool-specific history.

Because both exist, a tool that clears only one leaves the other intact — and viewers differ in which they display. A document can therefore look clean in one application and show the author's name in another, which is precisely the failure mode you were trying to avoid.

Handling both together is the only way to be confident. It is also why checking the result in a different viewer than the one you cleaned it with is a reasonable habit for anything sensitive.

Stripping everything is right when anonymity matters: an anonymous submission, a document from a template, a whistleblower disclosure, anything where the connection between the file and a person should not exist.

For published documents, thoughtful values usually beat empty ones. A filled-in title is what search engines display and what document management systems index, so an empty one makes your report harder to find and less professional in a list of downloads. Setting the author to your organisation rather than an individual gives you attribution without exposing a person.

A practical middle path covers most cases: keep title and subject, set author to the organisation, and clear creator and producer, which reveal your software stack and nothing useful to anyone else.

Metadata is one of several ways a document says more than its pages. Text hidden under a black rectangle is still in the file and is recoverable by copying — Redact PDF destroys it properly, and no amount of metadata cleaning touches it.

Images embedded from a phone carry their own EXIF data inside the image stream, including the coordinates where the photograph was taken. A PDF built from photographs can therefore carry location data that document-level cleaning does not reach; strip it from the images before building the PDF.

And the filename travels with the file while belonging to none of it. For anything published, reviewing the name is as worthwhile as reviewing the properties.