PDF Metadata Editor

Document Metadata: The Information You Didn't Mean to Share

Every PDF carries hidden information about itself, who authored it, with what software, when, and where in the world (via timezone offset). This metadata is invisible in normal viewing but exposed by anyone who opens the file's properties, runs forensic tools, or uses metadata-aware search. For most documents this doesn't matter; for documents going outside your organization, to clients, regulators, opposing counsel, or the public, scrubbing it is a small habit that prevents real disclosures.

This editor lets you see what's there and decide what stays.

What's Actually in the Metadata

A PDF has two metadata stores that often duplicate each other:

1. Document Information Dictionary (legacy). A simple key-value structure in the PDF's trailer:

• Title
• Author
• Subject
• Keywords
• Creator (the authoring app)
• Producer (the PDF-writing library)
• CreationDate
• ModDate

These are strings and dates. Every PDF has them; many tools auto-populate them from your system (Author defaults to your computer's full name in Word; Creator/Producer reveal your software stack).

2. XMP Metadata (modern, PDF 1.4+). An XML packet (Adobe's Extensible Metadata Platform) embedded as a stream. Same fields as above plus extensions for camera EXIF data, GPS coordinates, IPTC press tags, document history, custom schemas.

The two should match. They don't always, sloppy tools update one but not the other. Some forensic tools cross-reference them to detect tampering. This editor updates both consistently.

Real-World Metadata Leaks

The famous case: in 2017, a Microsoft Word document leaked metadata revealing the document had been edited at the same offices as a known Russian intelligence operation. Same scenario plays out at smaller scale routinely:

• Law firms filing briefs with "Author: Senior Partner's Computer" giving away document provenance.
• Companies sending "anonymous" contractor proposals with metadata revealing the actual author.
• Academic researchers sending blind-review submissions with their name in the Author field.
• Journalists sharing leaked documents where the original source's name is embedded.

The fix is always the same: scrub metadata before sharing. The cost is 30 seconds; the cost of NOT doing it can be a career.

Editing Patterns

Pattern 1: Strip everything before public distribution. Set all fields to empty or generic values. The recipient sees a "blank" PDF metadata profile. Useful for anonymous publications, leaked documents, redacted court records.

Pattern 2: Sanitize per privacy policy. Keep useful fields (Title, Subject) but strip identifying ones (Author, Creator, CreationDate). Common pattern for client deliverables, the document is identifiable as yours but doesn't reveal your internal tooling.

Pattern 3: Replace with deliberate values. For archival, set canonical Title/Author/Subject/Keywords that downstream search systems can use. PDF/A workflows require this.

Pattern 4: Read-only audit. Just view metadata to verify what's there before sharing. Common before sending sensitive documents.

Beyond the Editor: What's NOT Scrubbed

This tool handles the document-level metadata stores. Other identifying information lurks in PDFs:

• Font names with subset prefixes. Embedded fonts often have names like ABCDEF+ArialMT where ABCDEF is a subset ID. The ID is sometimes random per session; sometimes derived from the user. Forensic tools can correlate subset IDs across documents to link them to a common author.
• Image EXIF data. Photos embedded in the PDF retain their camera EXIF, model, serial, GPS, capture timestamp. PDF tools often pass this through unchanged.
• Annotation authors. Comments and form fields have author attributes separate from document metadata.
• Form field history. Some PDFs retain history of previous values.
• Optional content groups (OCGs). Layers that may have author attributes.
• JavaScript actions. Document-level JavaScript can reveal authoring tool.
• The content itself. Writing style, typos, internal references.

For high-stakes anonymization (whistleblowing, witness protection), use Adobe Acrobat's full "Sanitize Document" feature plus a content review. Don't rely on metadata-stripping alone.

Forensic Implications

Investigators routinely use PDF metadata as evidence. CreationDate proves "this document existed by X" or "this document was modified after X". Producer reveals what software ecosystem was involved. Multiple PDFs sharing a Producer signature can be tied to the same workflow.

This isn't paranoia, it's standard practice in digital forensics. The 2007 Iran nuclear documents, the 2016 Panama Papers, countless corporate investigations have hinged on metadata correlations.

If you're aware of forensic exposure, scrub before sharing. If you need to preserve metadata for evidentiary purposes (litigation hold, archival), don't edit, copy the original first.

PDF/A and Metadata

PDF/A (the archival subset of PDF) requires specific metadata:

• Title must not be empty.
• Conforming XMP metadata with PDF/A identification.
• No encrypted streams.
• Embedded fonts (subset is fine).
• No JavaScript or external references.

This editor produces metadata-edit output that's not PDF/A by default. For PDF/A workflows, edit metadata first, then convert to PDF/A with a dedicated tool (veraPDF for validation, Ghostscript or Acrobat for conversion).

Privacy

pdf-lib runs entirely in your browser. The file is read, parsed, modified, and re-serialized in JavaScript memory. Open DevTools Network during edits: zero outbound requests. Documents containing privileged information, confidential contracts, internal memos, regulatory filings, never leave the tab.