When Technology Democratises Deception

In late 2024, a woman walked into a Thai restaurant in Singapore's Katong Square, ordered nearly S$200 worth of food, and paid with PayNow. The staff verified the screenshot showing the transaction. Everything looked legitimate. Days later, during routine reconciliation, the restaurant discovered the payment never arrived. The woman had used a simple editing app to manipulate the PayNow receipt.

This wasn't an isolated incident. Between 2022 and 2024, similar schemes defrauded Singapore restaurants. One perpetrator made more than 35 fraudulent orders amounting to over $3,891.75 from a single restaurant over 15 months. Between 2022 and 2025, she placed food orders totalling more than $9,000 at two restaurants by using fake PayNow screenshots. Her method was disturbingly simple: transfer money to herself via PayNow, screenshot the transaction, use an editing app to change the name of the payee, and send the doctored image as proof of payment.

What makes these cases particularly instructive is not their sophistication but their simplicity. No elaborate hacking. No complex social engineering. Just readily available photo editing apps, tools that have been accessible for years, and the trust businesses place in visual verification. The fundamental shift isn't that AI made document manipulation possible; it's that AI has speeded up the scale and speed at which fraudsters can operate, while simultaneously making detection exponentially more difficult.

The Collapse of Surface-Level Verification

Assurance providers increasingly confront a fundamental verification problem when traditional market evidence is unavailable. As explored in our analysis of fair market value in illiquid markets, the collapse of observable transactions forces a reliance on internal models that require independent behavioural validation to remain credible.

The Singapore PayNow frauds expose a critical vulnerability in modern financial verification: we have optimised for efficiency at the expense of authenticity. Traditional controls like three-way matching, comparing purchase orders, goods receipts, and invoices, were designed for a paper-based world where creating convincing forgeries required specialised skills and equipment. Today these same controls falter because they rely fundamentally on document appearance rather than transaction substance.

According to the 2025 Identity Fraud Report, digital document forgeries surged 244% year-over-year between 2023 and 2024. The report documented that deepfake attacks now occur every five minutes globally, while fraudsters increasingly use generative AI tools to create sophisticated forgeries that pass traditional verification checks. The financial services sector bore the brunt of this shift, with cryptocurrency platforms experiencing fraud attempt rates of 9.5%, nearly double that of any other industry.

What we're witnessing is an asymmetry of capability. Fraud prevention tools improve incrementally through better scanning and pattern recognition. Meanwhile, fraud creation tools, powered by accessible AI, improve exponentially. A recent systematic review of AI-based identity fraud detection methods published in 2024 highlighted this challenge: "The use of Artificial Intelligence enabled deep fake technologies has significantly increased the complexity of identity fraud. Fraudsters may use these technologies to create highly sophisticated counterfeit personal identification documents, photos, and videos."

Three-way matching, once considered a gold standard for fraud prevention in accounts payable, now represents a false sense of security. The procedure verifies that three documents align, but if all three documents can be fabricated with editing software, the matching becomes meaningless. The control checks for consistency, not authenticity. It's the difference between confirming that three witnesses tell the same story versus confirming that the story is actually true.

This forces organisations back toward more labour-intensive substantive procedures. Vouching, the practice of tracing transactions backward through the accounting system to verify their occurrence, has regained prominence not because it's efficient, but because it's one of the few methods that can bypass fabricated documentation. Instead of checking whether a document looks right, vouching verifies whether the underlying economic event actually occurred.

The burden this places on assurance professionals is significant. Research on audit procedures in the digital era emphasises that "traditionally, auditors relied on manual procedures and sample-based testing, which often limited the scope and depth of their analysis." The return to more extensive vouching procedures represents a step backward in operational efficiency, consuming resources that could otherwise drive business value. For auditors serving business clients with lean finance teams, this burden becomes particularly acute: manual procedures that were once reasonable now strain both professional capacity and client relationships.

Moreover, the conventional audit approach creates temporal lag. Monthly or quarterly reconciliations mean fraud can persist for weeks before detection. By the time the restaurant discovered the fraudulent PayNow payments, the perpetrator had moved on to other victims. The traditional audit cycle, plan, execute, report, is fundamentally misaligned with the real-time nature of digital fraud.

From Document Checking to Behavioural Intelligence

The failure of document-centric verification points toward a necessary evolution: shifting from checking what transactions look like to understanding what transactions reveal about underlying behaviour. This is where consumption and behavioural analytics fundamentally differ from traditional audit procedures.

Consider the PayNow fraud cases through a behavioural lens. The perpetrator didn't just forge documents; she exhibited consumption patterns that deviated meaningfully from legitimate customer behaviour. She ordered from the same restaurant over 35 times in 15 months, always using the same payment method, always for significant amounts. These patterns, viewed individually, might appear normal. Viewed collectively and compared against broader consumption benchmarks, they represent statistical anomalies worth investigating.

Forensic accounting research increasingly emphasises this behavioural approach. A 2024 study on forensic accounting and fraud detection noted that advanced analytics can 'identify patterns, trends, and anomalies that might be missed with traditional methods' by processing vast amounts of transactional data to detect deviations from expected behaviour. The key advantage is that behavioural analytics don't depend on document authenticity; they analyse the substance of economic activity. Recent research in banking demonstrates this shift: studies show that the analysis of spending patterns and transaction data provides a nuanced understanding of customer preferences and financial behaviours, enabling institutions to detect anomalies independent of document verification.

This approach aligns with Seventwos' core thesis: that sophisticated behavioural intelligence provides a new form of document trust that visual verification cannot. When you understand how similar businesses typically transact, how consumption patterns normally evolve, and what legitimate customer behaviour looks like at scale, you can verify whether documents reflect genuine economic activity, even when the documents themselves appear perfect. It's not about ignoring the documents; it's about having an independent way to verify their authenticity.

The distinction matters because it changes the nature of audit work. Research on internal auditing's role in fraud detection confirms that "the use of data analysis techniques has been linked to increased efficiency, expanded audit coverage, and enhanced audit quality." Technology enables a risk-based approach, enhancing detection of anomalies and red flags through methodologies like continuous auditing. Instead of spending hours vouching invoices back to bank statements, assurance professionals can flag statistically unusual patterns for investigation. Instead of treating every transaction skeptically, they can prioritise scrutiny based on behavioural risk. The burden shifts from exhaustive documentation review to intelligent pattern recognition.

For assurance professionals serving business clients, this represents not just efficiency but a fundamental capability gap. Your clients cannot realistically sustain the level of substantive testing that burdensome procedures now require. You need a way to establish document trust without manually vouching every transaction. You need systems that can flag when consumption patterns don't align with the documents being presented, when the behavioural reality contradicts the documentary claims.

This is precisely the gap Seventwos addresses. By combining AI-powered behavioural analytics with consumption-pattern intelligence, we provide trust analytics: a way to verify document authenticity through economic substance rather than visual appearance. Our technology assesses whether documented transactions align with legitimate business behaviour, eliminating the need for burdensome manual vouching of every client transaction. It does not replace professional judgment; instead, it adds an independent verification layer that documents alone can no longer provide.

The broader implication extends beyond just flagging suspicious documents to rebuilding trust in documentation itself. When you can verify your business clients' documents against their consumption patterns and market benchmarks, you restore a level of assurance that visual verification has lost. The same analytics that detect fabricated documents also confirm legitimate ones. What starts as fraud detection becomes document validation: the ability to say with confidence which documents reflect genuine economic activity and which do not.

Towards Intelligent Assurance

The PayNow frauds in Singapore serve as a microcosm of a larger challenge facing the assurance profession. As AI makes fraud easier to commit, traditional verification procedures become simultaneously more important and less effective. The solution isn't simply doing more of what we've always done; it's fundamentally reconceiving how we establish transaction authenticity.

The path forward requires moving beyond document appearance to economic substance, beyond periodic checking to continuous monitoring, and beyond isolated transaction review to pattern-based intelligence. It requires accepting that in a world where documents can be fabricated perfectly, the documents themselves can no longer serve as primary evidence. Instead, evidence must come from behavioural consistency, consumption logic, and market-relative norms.

For assurance professionals, this represents both challenge and opportunity. The challenge is acknowledging that methods which served well for decades now require fundamental rethinking. The opportunity is using technology not to automate existing procedures but to enable entirely new forms of assurance: assurance grounded in behavioural intelligence rather than documentary evidence.

Seventwos exists to make this transition accessible to assurance professionals serving clients who lack enterprise-scale resources. By democratising sophisticated behavioural and consumption analytics, we're providing trust analytics that work even when the documents themselves can be fabricated perfectly. The burden of audit procedures need not rest solely on manual review; intelligent systems can lighten it by verifying documents through behavioural substance rather than superficial appearance.

The mission of accountancy professionals has always been the faithful representation of economic phenomena. In an era where documents can lie flawlessly, that mission requires us to look beyond the paperwork to the phenomena themselves: the real patterns of consumption, behaviour, and economic activity that documents are meant to reflect. Seventwos empowers you, the assurance provider, to achieve this on behalf of those you serve:

Lighten your burden with Seventwos.