When a PDF Claims a Respectable Origin: A Producer Field Case Study

This article is a snapshot – content was accurate as of July 2026 (code examples tested against the API as of June 2026). The product evolves actively; specific counts, examples, and detection rules may have changed since publication – see the changelog for the current state.
A fraud reviewer opens a PDF bank statement. One of the first things a manual check looks at is the Producer field, the line of metadata that names the software that wrote the file. This one names a well-known software vendor. To a human, and to most lightweight metadata checks, that reads as reassuring: it is professional software, the kind a bank’s back office or a law firm would use. The reviewer moves on.
This is an illustration, not a report on one customer’s file. It is built to show a pattern we see in the kind of document HTPBE? is asked to check: a file that claims a respectable origin, and a reviewer who takes the claim at face value.
A claim is not evidence
The Producer field is text stored inside the file. It is the file describing itself, and a description of that kind is cheap to change. A check that stops at ‘the producer names a famous vendor, looks fine’ is checking the one thing a person preparing a doctored document controls completely.
There is a second problem. The advice to distrust documents whose producer names a consumer tool is public, and the people it is aimed at can read it too. A reviewer who rejects the file that names a consumer tool and accepts the file that names a famous vendor is rewarding the more careful forger. You can read the field yourself in any PDF with the free PDF metadata viewer, and it is worth doing. Just do not treat what you read as a finding. The general case for why a viewer cannot settle this is in metadata viewer vs file-history check.
What the check reported
Imagine the statement above goes through HTPBE?. The result comes back modified, with one named check:
{
"status": "modified",
"modification_markers": ["HTPBE_PRODUCER_IDENTITY_FORGED"]
}The published label of that check reads: ‘Evidence the document’s declared institutional generator was spoofed by a third-party tool to disguise its origin.’ The label describes the pattern the check is built to catch. The result itself is a finding about the file, not about anyone’s intent, and it does not say who did it.
To the human eye the file looked trustworthy because it named a vendor everyone knows. That is the point of running a check rather than reading the metadata yourself, and the copy worth comparing against is one requested directly from the issuer. A result of intact, for its part, means only that no evidence of change was found. It is not proof that a document is genuine.
What the reviewer should do next
A modified result is a strong finding. It is not a verdict on a person.
- Hold the document. Do not decide on it, and do not decide against the sender on the strength of a tool result alone.
- Ask the issuer for a fresh copy. Request it directly from the bank, employer or agency named on the document, not from the person who sent it.
- Keep a human in the loop. Every API result carries a
usage_cautionobject for this reason: the result describes the file, not anyone’s intent. What to do after a modified result covers the conversation with the sender.
Files are edited and re-saved for ordinary reasons all the time. The finding is that the file shows evidence of change after it was created. Whether that is fraud, a careless export or a well-meant correction is for the reviewer and the issuer to establish.
What HTPBE? knows and does not publish
HTPBE? knows the fingerprints of the major PDF generators and editors and uses that knowledge as one input among many. It does not publish which ones, or what a particular result rests on.
That is a deliberate choice. A written account of what separates a genuine file from a spoofed one is not neutral information. For the honest reader it is trivia. For someone preparing a doctored document it is a specification. So we describe each check by what it finds and never by how it finds it. The list of checks, in outcome terms, is on the how it works page.
For the wider picture, see how to check if a PDF has been edited.
What a result does not prove
- No check is a guarantee. No method can promise to catch every forgery, and we do not claim it.
- Not every file can be judged. Some files come back
inconclusiverather thanintact, because the file alone does not settle the question either way.inconclusiveis neither an accusation nor a tool failure. If the result isinconclusiveand the decision matters, ask the issuer. - Not identity checking. HTPBE? examines the file, not the person. It does not read the content to decide whether an account holder exists or a balance is plausible, and it is not a KYC platform. It sits next to your identity and risk checks and does not replace them.
Running it in your own process
Submit a document and read the result. You do not request any particular check; every submission is run through all of them, which are listed on the how it works page.
curl -X POST https://api.htpbe.tech/v1/analyze \
-H "Authorization: Bearer $HTPBE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"url": "https://example.com/statement.pdf"}'That returns a check id:
curl https://api.htpbe.tech/v1/result/CHECK_ID \
-H "Authorization: Bearer $HTPBE_API_KEY"The request and response fields are described in the API documentation. A test key returns deterministic synthetic results, so you can build your review routing before sending real documents.
Teams that accept customer-supplied PDFs claiming to come from a bank, an employer or a government body can put this between the upload and the decision. The cost of a single check, and the one-off packs for running a batch before you integrate, are on the pricing page.