Fighting Fraud: How AI and Automation Combat this Scourge
We often talk about the cost and time reduction advantages companies realize in their accounts payable functions by implementing artificial intelligence (AI) and automation tools. But the ability to identify and combat fraud is a vital weapon in this technology toolbox.
Combatting fraud is an ongoing battle that every company experiences. The cost to businesses overall is huge and it is increasing:
- 78% of treasury and financial professionals surveyed in a 2018 Association for Financial Professionals (AFP) report experienced fraud in the prior year.
- 5% of all revenue is lost to occupational fraud annually, according to an Association of Certified Fraud Examiners (ACFE) report.
- $6.3 billion – that is the total loss to businesses in 2,410 occupational fraud cases cited in the same ACFE report cited above.
If you think this is something that only happens to smaller or less sophisticated enterprises, consider the case of Evaldas Rimasauskas, a Lithuanian man who pled guilty in March, 2019, of scamming more than $100 million from Facebook and Google by impersonating a company that both giants did business with. (Read the whole story.) The point is, companies of all sizes can be victims of those looking to defraud them.
How did Facebook and Google, two of the most technically equipped corporations fall victim to this? Rimasauskas sent invoices as email attachments, thereby still involving the human component. Emailed or snail-mailed invoices are definitely less secure than e-invoices and thus subject to fraud and error.
Victimization can be mitigated by P2P automation
In a perfect world, where all invoices and payments go through a fully automated procure-to-pay process with the correct rule engine, it is much more difficult to fall victim to fraud. However, if it happens it is much easier to detect by using a detailed audit trail to discover who has been making the fraud attempts and from where.
When Sarbanes-Oxley was passed in 2002, companies that may not have done so in the past, were forced to look at internal accounts payable controls. This has been a great driver for the adoption of P2P solutions. In the beginning, adoption was primarily among larger or international companies and corporations. That has been changing over the past ten years.
Unfortunately, for companies that have not implemented this automated process, or have different solutions that are not fully unified, fraud and errors are still common. This is due, once again, to receiving paper invoices that cannot be automatically matched against a purchase order and receipt of goods. The value of e-invoices is undeniable. But how can fraud be detected in companies that still need to scan invoices?
This is where AI can help
Artificial intelligence (AI) can analyze the unstructured documents that have been received and scanned and, thanks to new deep continuous learning solutions, extract all information automatically, thus relying less and less on human intervention.
This information can then be analyzed against whatever data is available. This is where AI helps to reconcile information; for example, finding the supplier from fuzzy names and addresses, or categorizing goods and services based on descriptions.
Automatic coding can then be applied and checked against rules and budgets, and lastly machine learning can help detect anomalies or fraud by associating a risk score to each transaction based on pattern detection (time, price, type of good, duplication, etc.).
The risk score can then be used either to simplify the validation workflow (no risk; no other approval than the strictly required one) or to make sure controllers or managers can be added to the loop to review the invoice before it gets paid.
As I’ve indicated throughout this posting, most fraud relies on repetitive, low-value, manual work, either internally or when using low-cost shared services centers to enter invoices and AP information. By using AI systems that do all these tasks, it is possible to get AP professionals back in the loop of having the proper insights to do much more rewarding tasks. When you let AI focus on detecting and mitigating fraud, your employees can focus on looking at the way money is being spent and creating a more engaging supplier relationship. And that, in the end, will lead to greater productivity and profitability.
See how you can reduce fraud and eliminate manual check runs.