How does AI provide an opportunity to reduce corporate fraud? 

Carolina Borzatto, Marketing Director for Latin America at SAP Concur, explains how intelligent expense management can reduce fraud and increase business productivity.

After salaries, corporate expenses represent the second operating cost for companies and are a crucial source of internal fraud. According to data from the Association of Certified Fraud Examiners (ACFE), they produce a loss equivalent to 5% of total revenues.

Corporate fraud is a problem that crosses the entire company. Most frauds are committed by employees or managers (41% and 35% of incidents, respectively), and critical losses account for 20% of actions taken by top executives. Frauds are also found in numerous areas, mainly operations (15%), accounting (14%) and sales (12%).

The variety of alternatives found by those committing fraud is large and growing daily. Some examples are duplicate meal charges, expenses without receipts, cash payments without vouchers, purchases for family or friends passed on as business costs and counterfeit tickets.

But today, companies have fundamental tools to contain the escalation and even anticipate future fraudulent strategies such as Artificial Intelligence (AI) and Machine Learning.

These technologies now find patterns in the data and detect any anomalies using algorithms based on mathematical models.

With recent advances, AI can interpret dates, vendor names, quantities and any other data in documents issued in different languages and countries, and cross-reference this information with corporate policies, authorizations, exceptions that could have benefited or even market averages.

For example, if each employee traveling to a destination spends a certain amount of money per day, any employee at that same destination who needs twice as much money will automatically trigger an alarm. But beyond this, AI can find much more subtle signs.

Any anomaly, no matter how small, becomes a ‘clue’ that it will activate the technology until it finds an acceptable resolution.

A fact is that fraud detection happens in near real-time. The ACFE study mentioned above pointed out that a typical case takes no less than 14 months to be detected and, during that time, produces very significant incremental losses.

When a fraud attempt is successful, in many cases, it becomes the incentive for trying it a second time.

The market is becoming aware of the opportunity presented by control over spending and early identification of potential fraud. Data from SAP confirms that approximately 41% of companies worldwide have increased their global budget for anti-fraud programs and 48% have invested in technology as a timely way to tackle the problem.

Companies were already working on this, but now technology can contribute to solving each case earlier and generate a better knowledge of their own and the market indicators more efficiently – not only to avoid losses but also to ensure savings.