Complexity & Expense Controls. Why the Finance Function has an Expenses Challenge.

Corporates are faced with a barrage of developments that make it harder to manage expenses successfully. This includes the rise of payment types, changes to working patterns, growth in regulation and demands from the business for faster closing. For many companies, expense controls are now complex and unwieldy.

How has this happened? Next week we’ll take a look at what can be done to improve the expense controls process by expanding the usual expense control to accounting controls that use AI. But first, let’s understand what’s making expense controls complex, then we can look at how to simplify the system.

A rise in payment types.

There has been a rise in small recurring payments such as subscriptions and licences. Global entities will have thousands of expenses transactions a day, many spending channels, cross-border expenses and multiple ERP systems in play too. It is a huge burden to check everything in detail. So managerial approval for expenses has been based on ‘does it look right?’ rather than automated, continuous controls. Added to this, sampling audits generally only review 5% to 10% of claims, so employees know there is a high chance that their expenses will pass unchecked.

Changes in working patterns.

Workers are not only in global offices, they are also working from home or in a hybrid set-up. This means that the traditional model of centralised purchasing is less effective and spending authority is dispersed across a wide range of locations and cultures. This increases manual approval burdens when documentation does not align across the organisation, making it harder to detect unusual transactions.

Growth in regulation & governance requirements.

The UK government has been tightening its regulation environment. The Bribery Act, and the more recent Economic Crime and Corporate Transparency Act (ECCTA) place more responsibility on organisations and their senior teams. The ‘failure to prevent fraud’ clause means that spending must not only be approved, but documented, auditable and compliant.

Demands for faster closing.

Businesses want to make confident decisions and rely on accurate data to do so. One area under
focus is the closing process. Senior teams are demanding greater, up-to-date visibility into cash flow
and spending. Meanwhile, finance teams are still grappling with disconnected files, a lack of co-
ordination between teams and late information submissions. These challenges make it difficult to
ensure expenses are accurate while also closing books quickly.

Added to this, AI makes fraud easier.

Recent research found that over one quarter of UK employees (29%) admit to using AI to generate or manipulate expense receipts to top up their salary. And over a third (32%) used employer-funded AI tools to do it. An SAP Concur global report, found that ‘67% of chief financial officers, 78% of travel managers, and 55% of business travelers’ are aware of this and believe the falsifying of expenses is likely in their organisation.

Expense control complexity is affected by how AI makes it easy to fake or amend receipts, invoices
and other expenses claims. For example:

• Receipts. Here the type and date might be real, but the image is generated by AI.
• Invoices. In these cases, a genuine document is modified by AI, so that the amount, date or beneficiary are false. This leaves the structure of the document in place and so can be hard to detect.
• Hybrid. Fraudulent claims use AI to create booking confirmations, mileage logs etc. to support a fictitious expense. When it is systemic, this over-documentation can be the giveaway; it becomes a signal that something is not right.

The receipt may be real, but the employee can use an AI editing tool to amend it. So an £80 dinner becomes a £180 dinner. For the team approving the expense, the traditional checks work: the logo is right, the format is the same, even the VAT number is correct. If the closing process is largely manual, the ‘does it look right’ question is outsmarted by an employee with AI. The question now needs to be ‘does it reflect typical spending patterns’, which is much better answered by a continuous, automated process.

Complexity comes at a high cost to the business.

Expenses fraud is often justified by the idea that everyone does it, it’s not a huge amount, and the company can afford it anyway. For workers facing a cost-of-living crisis, generating false expenses is an easy, often undetectable, way of boosting their income. It is a type of case that shows the three elements that can lead to fraud: motivation (the company can afford it, plus pressure from daily costs), rationalisation (everyone does it) and opportunity (AI makes it easy). UK Finance Directors know that this is already happening. The question is at what scale?

So the cost to the business is significant and likely to rise. The ACFE 2026 report estimates that globally, businesses lose 5% of revenue to occupational fraud. It also says the share of fraud cases that are expense-related are around 13% reported occupational fraud cases 3 . Based on these figures, a company turning over £10m could be losing around £65,000 in expenses fraud a year. The arrival of generative AI makes expenses fraud easier, which is expected to mean it rises at scale.

In many organisations, data volume has now out-paced the traditional manual controls the finance team uses. These may be online (Excel etc), but the data is entered and approval given manually. Work in this area has identified five blind spots in expenses control, which repeatedly emerge in diagnostics carried out in mid-sized companies:

• Appearance. Controls that rely on how a document looks and not on structured data are
  outdated. Seeing is not believing. It is no longer acceptable to assume that a document that
  “looks” authentic is authentic.
• External checking. Checks like company registration numbers and VAT should be verified
  automatically. This can be done via free public APIs.
• Formats. Motivation increases if the system will accept unverifiable formats. Tolerating
  photos and scans for example, is outdated in a world where image generation is easy.
• Siloed systems. Tools should integrate corporate cards, booking systems or calendars.
  Insisting that all expense spending happens on the company card gives the accounts team
  direct transaction data.
• Absence of deterrence. When employees see controls as superficial, fraud becomes a
  rational option. Without a modern deterrence, the opportunity, and the justification for
  fraud are increased.

Of course, expense fraud has existed long before AI made it easy and globalisation made it harder to track. Employees have tweaked their claims by inflating the cost of legitimate business expenses, claiming for the same thing multiple times, also known as ‘double dipping’, or just using a few extra blank taxi receipts from a well-tipped cabbie.

It’s the complexity that is challenging for finance teams today. Regulation, data volume and AI tools make their current systems unfit-for-purpose. Finance Directors know that the opportunity for fraud has been made easier by AI tools, coupled with existing controls that have failed to keep up. That said, they are also looking to state-of-the-art controls, that incorporate AI, to bring their systems up- to-speed and meet the challenge of modern fraudulent activity.

The challenge and complexity has never been greater. The solution could be in the threat, and just
around the corner.