Page 15 - CMA Journal (Jul-Aug 2026)

F oc us S ec t ion

Learning What is Not Normal Connecting Physical Goods with Financial The business should establish anti-fraud steering
Figure 2: The Procurement Cycle
Machine learning can identify unusual behavior even if Transactions committees. The members of such committee should
STAGE 1 FRAUD WARNINGS the exact fraud pattern has never been recorded The Blockchain and IoT can together e ectively be from various departments across the organization
such as nance, operations, risk, IT and Internal Audit.
[1] VENDOR ONBOARDING • New suppliers with unclear/hidden ownership structures before. Traditional systems normally depend on prevent ghost transactions and issuance of fake This diversi ed representation will bring di erent
prede ned rules such as agging the transactions
Sourcing and registering new suppliers • Supplier registered at same address as an employee above a certain amount. The fraudsters can invoices. For example, a seller dispatches 10,000 perspectives on fraud risk evaluation and prevention.
• meters of fabric to a garment factory. Suppose, the
immediately learn these rules and make plans to goods were never delivered to the factory. The Creating Safe Channels for Speaking Up
Introduction Tree: asset circumvents them. Unsupervised ML can learn documents like purchase order, delivery challan,
misappropriation and abnormal behavior and identify deviations from the inward gate pass and invoices were fake because of all It is easier for an employee to notice the unusual
Corporate fraud is often associated with sophisticated corruption. normal pattern. This is highly useful when methods of behavior of his or her colleague. Any wrong such as
cyberattacks, data breaches, and hackers in ltrating STAGE 2 FRAUD WARNINGS fraud are changing and evolving. For example, Master relevant persons colluded. giving favor to a particular vendor, posting of fake
organizational systems. While cyberfraud can pose One of the most [2] REQUISITION & P.O. • 'Split' contracts to avoid higher approval limits are always below the approval limits. Accounts The accounting system might have recorded the invoices or suspicious phone calls of a colleague can be
serious and potentially devastating risks to businesses, common forms of asset • Bid-rigging or sole-supplier card, a global payment network uses AI, ML and Department pays immediately. Moreover, the purchase of goods even though these were never easily sensed by other colleagues. However,
procurement fraud remains one of the most prevalent misappropriation is a • Manual override without documented business need anomaly detection software to identify transactions supplier's registration records are linked to another delivered. employees hesitate to report even sometimes very
forms of corporate fraud. According to the global billing scheme. Under which deviate from normal customer behavior. Since employee's personal business entity. Although, all obvious signs. They don’t want to risk their careers for
survey ndings published in the ACCA Policy & Insights this scheme, a business anomaly detection does not rely only on previously these events seem harmless in isolation. But, an AI However, if the IoT have been employed, the digital fear of retaliation. Moreover, if unproven after
report, “Combatting Fraud in a Perfect Storm,” is deceived to make marked fraud examples, it can identify new fraud system can analyze the hidden relationships and nd mirror of the physical movement of the goods would investigation, they would be regarded as
procurement fraud has a 34% global prevalence rate payments for STAGE 3 FRAUD WARNINGS patterns which never happened before. out the fraud scheme. have been recorded automatically. Additionally, troublemakers.
and a 35% materiality rating, highlighting its unprovided services or [3] GOODS RECEIVING • Inventory levels rising without corresponding sales growth Natural Language Processing: Reading Between blockchain could have saved the information in a
signi cant impact on organizations. unsupplied goods. Verifying physical delivery of goods • goods and services the Lines Blockchain and IoT: Building a Trusted tamper-resistant record. This record can be veri ed The business should establish channels that are
Procurement fraud can occur at various stages of the Normally, the ghost • Deliveries accepted without matching POs The clue for fraud is not merely hidden in numbers. Supply Chain with the physical record and in case of tampering, con dential and safe for the employees. A formal
fraud would be detected.
purchasing and supply chain process, from supplier vendors or shell This can be found out in emails, letters, agreements Businesses have become complex. A transaction whistleblowing policy should be implemented under
selection and tendering to contracting, invoicing, and companies are used to execute such schemes. Another and other written communications. Natural Language normally involves many employees from multiple Creating a Culture of Integrity which employees should be able to raise their
payment. Given that procurement is fundamental to way is using the Company’s credit card or purchasing Processing (NLP) can help to detect fraud in these locations interacting with various suppliers while The role of technology is very important in preventing concerns on more than one channel but anonymously.
acquiring the goods and services required for channel to buy items for personal use. STAGE 4 FRAUD WARNINGS sources. NLP can be used to analyze large amounts of engaging outsourced logistics. This is also done fraud, still the human factor cannot be ignored in fraud Visible Action
day-to-day operations, weaknesses in procurement The corruption is the second most important category. [4] INVOICING & BILLING • Round-number invoice amounts (e.g., Rs. 10,000, Rs. 5,000) unstructured data to identify unusual words, themes, through scores of digital systems. It makes traditional detection. The businesses should promote the culture
controls can expose organizations to nancial losses, According to the ACFE, corruption occurs when an V • Invoice amount right below control limits security systems unreliable. Therefore, the businesses of ethics and integrity among employees. The ethical culture can strengthen only if the outcome
reputational damage, operational disruptions, and employee abuses his position or in uence for personal • Increase in wire transfers requests to new accounts patterns and relationship e.g. unclear and evasive have to install checks on devices, individuals and of investigation is communicated keeping all the
regulatory risks. Consequently, strengthening words such as “usually”, “normally”, "perhaps", systems along the way as transaction or event Shared Accountability necessary information con dential. The organizations
procurement governance, transparency, internal bene t, in breach of his duty. In procurement, this "generally". These words may provide a sign that the proceeds. Blockchain and Internet of Things (IoT) are Accounts, IT and Internal Audit cannot solely be held may publish and circulate newsletters on “speak-up
controls, and oversight is essential for organizations often involves manipulating the supplier selection person communicating about a transaction is trying to helpful and create a “digital mirror” of what is taking responsible for fraud prevention. Since fraud can outcomes” preferably on quarterly basis. These should
seeking to safeguard resources and maintain integrity process through practices such as bid-rigging and STAGE 5 FRAUD WARNINGS mislead. This may not detect a fraud but provide a lead place physically. happen anywhere in the organization, fraud detection discuss the concerns raised, investigation done,
throughout their supply chains. kickbacks. • Requests for urgent payments to bypass internal controls to a suspicious scheme. The American Securities and and prevention are the shared responsibility of corrective actions taken, and lessons learned. The
Understanding the Procurement Fraud Why Procurement Fraud is Hard to [5] PAYMENT DISBURSEMENT • Exchange Commission (SEC) has realized the Creating a Digital Mirror of Physical Operations procurement, operations, IT, accounts, internal audit objective of this communication is how seriously the
importance of language and uses an NLP software
concerned raised were considered without naming the
The Blockchain and IoT are used to capture information
Procurement fraud is an occupational crime. Detect? Payments to completed accounts • Last-minute bank detail changes bypassing the independent which is designed to read between the lines for about goods as these are moved through the supply etc. These functions must regularly communicate and individuals involved. This should be done without
share information about possible fraud risks. It is worth
Association of Certi ed Fraud Examiners (ACFE) Unlike a cyberattack, which can trigger sudden check detecting the probability of fraud in the chain. For example, if a transporter installs telematics noting that fraud normally happens where systems discrimination ensuring that the same rules also apply
de nes occupational fraud as the use of one’s security alarms, procurement fraud is often di cult to communication of executives of the companies. devices or trackers in his trucks, he can have to senior and high-performing employees.
occupation for personal enrichment through the detect because it is designed to look normal. In Fraud AI and Machine Learning: The New already been processed. On the other hand, an AI From Finding Fraud to Connecting the Dots timestamped record of the location, and movement. and work of di erent departments meet with each Becoming a Hard Target
deliberate misuse or misapplication of the employing Analytics Using Descriptive, Predictive, and Social system can continuously analyze large volumes of This can be used to trace the goods. It may also be used other. In fact, these are fraud detection points;
organization’s resources. Procurement fraud generally Network Techniques by Bart Baesens, Véronique Van Smoke Alarm transactions in real-time processing. It helps to identify AI’s strength is not limited to nding an instance of to review a claim of a disgruntled client in case of a therefore, communication among these departments The fraud can never be eliminated fully because bad
falls into two main categories under the ACFE Fraud Vlasselaer, and Wouter Verbeke, this As discussed above, a procurement fraud can blend unusual patterns on the go. The unusual transaction fraud in isolation such as in a single transaction or a dispute about the delivery of goods happens. is vital. people lter through the organizational hierarchy.
behavior is compared with the camou age into normal business activities which makes it hard to times, repeated round-number transaction or other single account. It can connect many signals, which is Blockchain: Creating a Shared Version of the Truth Such people are also normally happened to be very
necessary, because modern fraud schemes depend on
techniques used by animals such as detect. Moreover, in this age of information abnormal patterns can easily be identi ed and agged. relationships between people, device, and accounts. creative, adaptive, and tech savvy. Still, a combination
chameleons and stick insects. Abdelrahim technology millions of transactions can take place at For example, Accenture has integrated an Therefore, an important analytical shift has happened While IoT is helpful in knowing what is happening of AI, blockchain, IoT and ethical culture can make an
Al Aqqad, in Fraud Analytics in Action, also very high speed without the use of a single piece of unsupervised AI-based anomaly detection model into physically, blockchain makes it possible to record and organization a hard target for the fraudsters. They go
discusses this “camou age e ect.” paper making human supervision di cult. Therefore, its corporate expense reporting system. This solution from simply " nding fraud" to "connecting the dots" save this information digitally in Trustworthy way. A away for easier targets when they realize that an
Fraudsters deliberately imitate normal businesses need tools that can sense small signs that continuously screens 100% of expenses to isolate through arti cial intelligence and Social Network shared blockchain ledger can be used to record organization is actively searching for anomalies and
business behavior so that their actions humans may overlook. In this regard, Arti cial statistical outliers i.e. unusual expense claims and Analysis (SNA). The complex entities and their information collected by sensors of IoT. The entries in asking the right questions.
blend in with legitimate employees and Intelligence (AI) and Machine Learning (ML) can play hidden patterns of non-compliance. This system has interactions are modelled as network graphs through the blockchain ledger cannot be changed without
routine transactions. The gure 2 provides a an important role. signi cantly reduced the auditors’ time and enabled SNA which shifts the analytical lens from individual authorization after these are properly recorded and
brief snapshot of the procurement cycle Detecting Anomalies in Real-Time them to utilize it for investigating the suspicious behaviors to overall relational structures. For example, validated. This highly reliable record can be shared
along with potentially fraudulent activities transactions agged by the AI. a procurement manager again and again awards with con dence among buyers, suppliers, transporters
that look business as usual. Under a traditional audit a sample of invoices and contracts to the same supplier. The supplier's invoices or any other party.
transactions is checked and veri ed after these have

ICMA’s Chartered Management Accountant, Jul-Aug 2026 13