How to Tell If a PDF Has Been Edited, and What Proves It
You can find clues that a PDF has been edited: modification dates, the software named in its properties, extra saved revisions, or differences from another copy. None proves anything on its own. Proof comes from a valid digital signature, a matching checksum against a trusted original, or the audit trail of the system that holds the record.
That distinction matters most when the document is evidence: a supplier's certificate of analysis, a signed protocol, a batch record extract or a contract. This article goes through the clues, explains why each can mislead, and then sets out what actually counts as proof.
Clue 1: the metadata
Most PDFs carry document metadata. ISO 32000-1 section 14.3 describes two places for it: the document information dictionary, which holds entries such as Title, Author, Creator, Producer, CreationDate and ModDate, and an XMP metadata stream, which PDF 1.4 introduced and the standard now prefers.
Open the file in View Metadata and look at:
- Creation and modification dates. A modification date well after the creation date shows the file was saved again later.
- Creator and Producer. Creator is usually the application the document was written in; Producer is the software that made the PDF. A certificate that claims to come from a laboratory system but names a consumer PDF editor as Producer deserves a question.
- PDF version and other fields. These tell you about the software history, not about content.
Why it is not proof. These are ordinary text fields. Many tools rewrite them on save, some clear them, and anyone can type in new values: our own Edit Metadata tool lets you set the Creator and Producer to any text you like. A clean, consistent set of metadata does not show a file is unedited, and an odd one does not show it was tampered with.
Clue 2: saved revisions inside the file
A PDF can be updated without being rewritten. ISO 32000-1 section 7.5.6 says that when a file is updated incrementally, the changes are appended to the end of the file, leaving its original contents intact, and each added section ends with its own end-of-file marker, %%EOF. A file that has been updated several times therefore contains several trailers, and earlier copies of changed objects can still be there.
You can look for this yourself: open a copy of the PDF in a plain text editor and search for %%EOF. More than one suggests later revisions.
Why it is not proof. Several innocent things produce extra sections. A linearised (fast web view) file has two cross-reference sections, and the first may also end with %%EOF. Each digital signature added to a document is saved as an incremental update. In the other direction, any tool that writes a complete new file leaves a single clean section and erases the history. One %%EOF means only that the last tool rewrote the file.
Clue 3: what the pages look like
Close inspection can reveal edits that metadata hides: one line in a slightly different font or size, a number that sits off the baseline, text that highlights separately from the words around it, or a scanned page where one figure can be selected as text while the rest of the page is an image. Zoom in to 400 percent and select text across the area you care about.
Why it is not proof. Mixed fonts and odd selection behaviour have ordinary causes, such as text pasted from another document, OCR applied to part of a page, or form fields. Treat what you see as a reason to check further.
Clue 4: compare it with another copy
If you have an earlier version, or a copy received through another channel, a comparison shows exactly what differs. Compare PDFs does this in your browser. It extracts the text of both files, pairs up pages by how similar their text is, compares them word by word and highlights added, deleted and changed words, and it also flags words whose font differs between the two files.
Know its limits before you rely on it:
- By default it hides differences in the top and bottom 12 percent of each page, treating them as headers and footers. Version numbers, dates and page references often live there, so switch the header and footer filter on.
- It compares text only. Changes to images, drawings, signatures drawn as images or colours are not detected.
- A scanned page with no text layer has no text to compare. Run OCR on both versions first, and remember OCR can misread.
Why it is not proof. A comparison tells you two files differ, not which one is authentic. It is only as good as the copy you compare against.
What actually proves a PDF has not changed
A valid digital signature
A certificate-based digital signature is designed for exactly this question. ISO 32000-1 section 12.8.1 says a signature stores a digest of the document's bytes; to verify, the digest is recomputed and compared, and differences indicate that modifications have been made since the document was signed. Open the file in a full PDF reader such as Adobe Acrobat Reader and check the signature panel.
Read the result carefully. A signature can be valid for the revision it signed while later changes have been appended, because an incremental update preserves the signed bytes. And a valid signature proves the content has not changed since signing; whether you trust the signer is a separate question about their certificate. A signature image pasted on the page proves neither.
A matching checksum against a trusted original
If a trusted copy, or a checksum of it, was recorded when the document was approved or received, you can recompute the checksum of the file in front of you and compare. In practice, changing even one byte gives a completely different SHA-256 checksum. The eCTD specification relies on this: ICH eCTD v3.2.2 asks for a checksum for each file so that the integrity of each file can be verified by comparing the submitted checksum with the computed one.
The audit trail of the system of record
In regulated work the PDF is often not the record itself. The FDA's data integrity guidance (December 2018) defines an audit trail as a secure, computer-generated, time-stamped electronic record that allows reconstruction of the course of events relating to the creation, modification or deletion of an electronic record. The MHRA's GxP data integrity guidance (March 2018, section 6.13) describes it as recording the "who, what, when and why" of each action. If a PDF is a report printed from a validated system, the question "was it edited?" is answered most reliably by comparing it with that system and its audit trail.
The same guidance explains when a PDF can stand in for the original. MHRA section 6.11.2 defines a true copy as one verified, by a dated signature or a validated process, to have the same information as the original, including the data that describe its context, content and structure.
If you suspect a document has been altered
- Do not edit or re-save the suspect file. Work on a copy, so the original bytes, including any earlier revisions, are preserved.
- Record a checksum of it now, so you can show later that the file you examined is the file you received.
- Gather the clues: metadata, revision markers, signature status and a comparison with any other copy.
- Ask the source. Request the document again through a controlled channel, or ask the issuer to confirm the values that matter.
- Escalate under your procedures. In GMP work that usually means a deviation or a supplier quality issue, handled by QA rather than settled by a tool.
For more on choosing and using PDF tools around regulated records, see our article on online PDF tools and GMP data integrity.
This article is general information, not legal or regulatory advice. Where a document's authenticity has legal or quality consequences, involve your QA, legal or DPO team.
Tools mentioned in this article
Frequently asked questions
Can the modification date show that a PDF was edited?
It shows that the file was saved again after it was created, but it is an ordinary text field that many tools rewrite or clear and anyone can change. Treat it as a clue, not as evidence.
How do I see earlier versions saved inside a PDF?
A PDF saved by incremental update keeps its earlier content and ends each saved section with a %%EOF marker. Searching a copy of the file in a text editor for that marker shows how many sections there are, though linearized and signed files also have more than one.
Does a valid digital signature prove a PDF was not edited?
It proves the signed bytes have not changed since signing. A reader may still report changes appended after the signature, and whether you trust the signer depends on their certificate.
What should I do if I think a supplier document was altered?
Keep the file exactly as received, record a checksum of it, compare it with any other copy, ask the issuer to confirm the key values through a trusted channel, and raise it under your quality procedures.
Sources
- ISO 32000-1:2008, Document management: Portable document format, Part 1: PDF 1.7 (free copy published by Adobe)
- FDA: Data Integrity and Compliance With Drug CGMP, Questions and Answers (December 2018)
- MHRA: GXP Data Integrity Guidance and Definitions, Revision 1 (March 2018)
- ICH eCTD Specification v3.2.2 (16 July 2008)
