Doctoring receipts to commit expenses fraud is not a new concept, but it’s never been easier. The existence of free AI tools means that anyone can now fabricate convincing evidence of expenses in seconds. This new wave of AI-generated expenses is slipping through checking systems and costing firms millions.

According to an October 2025 report in the Financial Times, the launch of OpenAI’s GPT‑4o model last year coincided with a sharp rise in falsified receipts. The FT reports that software provider AppZen reported fake AI receipts accounted for about 14% of fraudulent documents submitted in September, compared with none in 2024. In addition, Fintech group Ramp’s software flagged more than $1m (£759,868) in fraudulent invoices within 90 days.

Accessible generative AI has transformed document forgery. “Fraudsters are using generative AI to produce fake documentation such as pay slips, tax forms, receipts, invoices and bank statements which are then submitted to third parties for reimbursement or financial gain,” says Arun Chauhan, Director and Founder of Tenet Law, and a Business Fraud Alliance board member. “These fakes are now so realistic that they can be very difficult to detect.”

That realism extends far beyond just logos and formatting. With the right prompt, AI can generate the texture of thermal paper, simulate creases, and even reproduce the blur of a phone camera. Some clever fraudsters are also pairing fake receipts with AI‑generated voice or video deepfakes to impersonate senior staff and authorise false claims.

Use AI to catch AI

However, Ian Pay, head of data analytics and tech at ICAEW, says the tools that enable fraud can also help to stop it. “There’s increasing adoption of AI to be able to detect AI,” he says. “While some people think this is a bit like marking your own homework, the reality is that AI is often quite good at identifying imperceptible patterns in the way an image has been created.”

AI tools can now scan receipts for inconsistencies in lighting, texture, or alignment, as well as hidden digital fingerprints left during image generation. But technology is only part of the answer: “If you know what to look for, AI‑generated receipts are still fairly easy to spot,” Pay says. “Does it include all the usual information, dates, VAT numbers, store numbers? Does it look too perfect? Even when AI adds wear and tear, it often does so uniformly. Real receipts have quirks, smudges, and folds that are hard to replicate.”

Make sure the details correlate

The Business Fraud Alliance offers practical guidance that applies equally to AI‑driven expense scams. Red flags to watch for include receipts that don’t correlate with known locations, times, or employee activities, totals that repeatedly fall just below expense limits, absent or inconsistent VAT or transaction details and images that appear two‑dimensional or unnaturally lit.

A mix of logical checks like this and scrutiny is crucial. Many companies are already piloting verified digital receipts transmitted directly from merchants via secure APIs, bypassing the need for employee‑uploaded images altogether.

Check expenses like you would for AML

As with anti‑money‑laundering (AML) and know your customer (KYC) checks, the same principles apply – verify, contextualise and question. Fraudsters can move fast, but encouraging a culture of healthy curiosity and robust measures is still the best defence against even extremely convincing digital forgeries.

“Financial professionals can still carry out contextual verification of receipts by cross-referencing submitted receipts with employee diaries to check travel patterns and with historic expenses claims to check spending norms, says Senior Associate at Tenet Law and Business Fraud Alliance board member Esther Phillips. She adds: “Logic remains a key tool, but this does require allowing those doing the checking to have time to stand back and stress test and think about what they are being presented to approve.”

Pair automation with human experience

Ultimately, fraud prevention depends on combining automation with human intelligence. AI can detect anomalies at scale, but people can interpret them with context and experience. “You’re only really in trouble if your organisation has a culture of waving through claims,” says Pay. “AI receipts could be a problem, but systems for proper review should have been in place anyway. I’ve yet to see an AI‑generated receipt that would genuinely fool me.”