During one of our most recent Social Media campaigns, we shared some shocking statistics relating to expense fraud:
And if that wasn’t alarming enough, now it seems, the situation is getting worse.
As covered in a litany of publications including the Financial Times, financial professionals are reporting increased incidents of fraudulent receipts being passed through expense management systems than in previous years.
So what’s going on; why has expense fraud started to increase at a time where fraud detection systems are becoming ever more sophisticated?
The answer is simple: the mainstream use of AI tools like ChatGPT.
As recently as 2024, creating convincing fake expense receipts was not an easy task.
The images put through for claims had to appear absolutely genuine. This meant perfect natural lighting, wrinkles in the receipt paper, pinpoint accuracy in text alignment and typography, and business exact logo design.
This was not a task for your everyday employee. It required serious photo editing skills and a lot of patience and time, which generally speaking meant it wasn’t worth the effort.
However, this all changed with the advent of AI image generation.
It’s now becoming outstandingly easy for anyone to create impeccable fraudulent receipts with a few simple examples and prompts. Employees who would never have been able to commit such complex expense fraud can now do so in a matter of mere minutes.
Before we get to the good news, let's go over the bad news.
Without proper measures in place, it’s very very difficult to tell if an AI-generated expense receipt is indeed AI generated.
As a simple but poignant example, you may have an expense policy for meals during travel.
Using AI, an employee could generate a fake restaurant receipt and submit it for payment. Your business then pays out the value, let's say £100, to the employee to cover the expense. But, if they’ve not actually had the meal, and instead taken a home cooked dinner along for the journey, then they’ve just banked £100 at your expense.
This means employees can quite comfortably make fraudulent claims that sit well within your company’s expense policy and you’ll never know – unless you take aggressive steps to check their claims, which you may want to avoid given potential friction if the expense is, in fact, genuine.
And it’s this point that really matters.
Policing expense fraud can come at the cost of degrading relationships with genuinely hardworking, reliable and trustworthy employees.
While 50% of employees have committed expense fraud at least once, only 10% of people do this regularly, so most of your expenses are genuine expenses, and casting aspersions on everyone due to a few bad apples can absolutely be as costly as the cost of the fraud itself.
So how do you manage this rise in AI expense fraud without harming working relationships?
And now for the good news.
At ExpenseOnDemand, our expense management software offers an integrated credit card feed system, whereby all receipts can be reconciled with business cards set on file.
By simply aligning your expense policy to payments via credit cards, our system can then make sure that all payments are genuine and not AI-generated misdirects. Each and every payment will be automatically checked against your business card history.
Any claims made that aren’t allocated to a business card can then be investigated.
This can dramatically reduce fraudulent AI claims in more ways than one.
The first is that it can obviously flag payments that don’t match card history.
Following on from that, if you do find you have a receipt that isn’t matched to your connected business cards, you can reduce friction as you have justification for further investigations (rather than having to doubt all claims being made).
It can also act as a deterrent. If your employees know these checks are in place, they’ll be a lot less likely to try their luck with AI generated imagery.
But what if your company doesn’t provide linked expense cards? Fortunately, credit card reconciliation isn’t your only option for monitoring AI receipt fraud.
ExpenseOnDemand uses metadata captured at the point of claim submission to detect anomalies in expenses.
Each image or digital file uploaded to our app carries information like when and where the image was taken, what device captured it, and whether it’s been modified. Our platform leverages this metadata to detect inconsistencies or signs of manipulation that could indicate AI-generated or falsified receipts.
Our technology can look at items like mismatched timestamps, missing data (common in AI-generated images), or GPS coordinates inconsistent with claimed merchant locations.
This is certainly not bulletproof, as we all know employees don’t necessarily submit claims at point of purchase but, when combined with other approaches like our AI + Metadata Fusion – whereby our AI models learn the normal patterns across thousands of genuine receipts submitted by your team, then flag anything statistically unusual – it starts to become much more effective.
And, once again, because the system flags the anomaly, you can avoid internal friction by carrying out further investigations under the guise of following software protocol, rather than mistrust.
Fraud prevention is an ongoing battle and an ever-present part of expense management for finance teams. However, as emerging technologies create new challenges, our team at ExpenseOnDemand is ready to rise and meet them.
If you’re concerned about the potential impact of AI receipt fraud on your business, you can book a demo with us today and find out how we can work with your unique business to mitigate this growing risk.
Connect With Us
%201.png)
-transformed.png){kind=spacer}