Article 4 min

Artificial Intelligence, Hype and Financial Misconduct

AI for finding financial misconduct

September 25, 2018  


AI for finding financial misconduct

The compliance world, like business in general, is being inundated with “trends” and claims of “new technology” that inevitably (is or) will transform our economy.  There is no question that artificial intelligence holds great promise for compliance and our entire economy.

When you boil down artificial intelligence, it is built on two significant developments – the creation and collection of large data sets (sometimes called “assets”) and most importantly, advances in computer technology and power.  Over the last ten years, computer technology in terms of speed and computational capabilities has been revolutionary.

The developments in data has resulted in the new focus on “big data,” that is the ability of companies to generate, organize and store large amounts of data relating to its business operations.  Coupled with greater computer power and the ease with which such power is distributed among users (e.g. our smartphones), the trend is unmistakable – large amounts of data can be quickly mined or processed to identify a host of data sets, anomalies and other interesting inferences and results.

The revolution in this area is initially being advanced by financial institutions because they generate large numbers of transactions and have significant need to identify anomalies or potential fraud.  Credit card fraud is a massive problem that costs the industry as much as 5 percent of its annual revenues.

AI and Financial Misconduct

In the corporate compliance space, however, artificial intelligence or machine learning has had an impact in two areas: financial misappropriation and due diligence.

Internal audit faces an unending challenge in identifying, as quickly as possible, potential fraud or misappropriation of corporate resources.  It is easy to recognize the threat – money can be accessed and then used for illegal purposes such as paying bribes.  Here, artificial intelligence and data analytics can identify early potential anomalies.  Forensic accountants live for the moment when they are able to identify an “anomaly,” meaning a suspect transaction (or set of transactions.”

When reviewing a large amount of internal corporate data, finding anomalies is more difficult than locating the proverbial needle in a haystack.   Artificial intelligence – computer powered processing of large data sets – is now feasible for many companies to implement.  As a result, companies will be able to internally monitor financial transactions closer to the occurrence of the transaction.  Internal audit can now develop real-time monitoring programs.

As companies digitize their processes, they develop real audit trails in their enterprise management systems.  Applying artificial intelligence processes to monitoring and reviewing large collections of financial transaction data results in increased accuracy in identification of anomalies.

Artificial intelligence is not the magic bullet for financial accountants but it is and will be a valuable tool in every company’s arsenal in the battle against fraud.  Companies need to explore artificial intelligence capabilities, the application of such processes to their activities, and develop a strategy for implementing artificial intelligence on a cost-effective basis.  By 2021, most companies will incorporate artificial intelligence – machine learning into their internal applications.  Organizations that face large fraud risks will have greater need for implementing artificial intelligence solutions.

It is important to remember, however, that artificial intelligence fraud detection strategies should never be the only solution used by a company.  Relying on predictive fraud detection may be helpful but should never be the only means to detect suspect transactions.  Anomaly detection requires flexible approaches and strategies as fraudsters employ new or different schemes to carry out their crimes.

Fraud detection requires a holistic approach to harness artificial intelligence and detection strategies.  A multi-layered approach may be an effective way to balance a company’s strategy.  The challenge is to translate artificial intelligence into actionable intelligence.