Does a flag prove fraud?
No, and walk away from any tool that says yes. A flag is observable evidence: something concrete in the file your review can stand on. The call stays yours, and the report backs it.
If the report shows no findings, is the document genuine?
No, and we will never say it is. No findings means we found no observable indicators in that file, nothing more. A document can be fabricated cleanly, and a scan or photograph has no text to read, so most checks cannot run at all. The report distinguishes we looked and found nothing from we could not look. Treat it as one input to your review, never as a clearance.
A document with no saved history: is that a dead end?
No, and this is where the file-level layers earn their place. A document with no surviving history still faces the full workup — the figures and how they are printed, the identifiers, the typefaces, the way the rows are built, and how the file itself is assembled. Several of those read the file rather than the page, so they answer whether or not an earlier version survives.
Whose responsibility is the decision?
Yours, and nothing about using Fraud Finder changes that. It gives you observable facts and the record behind them. It does not verify a document, approve an applicant, or discharge the inquiries you are already expected to make. If you accept a document, that is your call and your responsibility.
Do you use AI for the document forensics?
No. None, anywhere. Every finding is deterministic: recomputed from the document itself, so the same file returns the same findings for anyone, every time. That matters under Australian law. From December 2026, privacy law requires transparency about computer programs that make or substantially contribute to decisions affecting individuals. Fraud Finder never puts you in that position: no AI model, no confidence score, no automated call. Facts you can verify, and a decision only a human makes.
What happens to my client's documents?
Nothing, and that is the point. The document is never sent to our servers and never stored, so there is nothing on our side to lose. Your clients trust you with their personal information, and every vendor you upload it to becomes part of your risk. When a vendor holding client files gets hacked, the fallout lands on the broker who sent them. Fraud Finder takes itself off that list: we could not read the documents if we wanted to, and no one can hack what was never stored.
Why use Fraud Finder instead of an AI model?
Because an AI model reads the page and gives you an opinion. Fraud Finder reads the file and gives you the facts. An AI model cannot recover the saves buried inside the file, and it can give a different answer every time you ask. Ask twice, get two opinions: that is not evidence. Then there is the law. A client's payslip is exactly the financial and identification information Australia's privacy regulator advises businesses not to enter into public AI chatbots, because it can end up stored on someone else's servers and used to train the next AI model. And distance does not dilute the duty: under Australian privacy law, when client information is sent to an overseas server, responsibility for what happens to it stays with the business that sent it. Fraud Finder never asks you to make that trade: the same findings for anyone, every time, and your client's document never leaves your control. A record you can put in front of an auditor.Source: Guidance on privacy and the use of commercially available AI products, OAIC, 21 October 2024
Do I need to install anything?
No. Fraud Finder runs on the machine you already use, in software you already have. Nothing for your IT policy to approve, nothing to update, nothing new to learn. Open it, drop in the document, print the report.