A milestone in the fight against corruption and corporate crime has been reached.
Artificial intelligence (AI) has been making a big difference in nearly every industry and is changing all of our lives on a personal level, but to date, it hasn’t curbed sophisticated white-collar crime. In this article, we will cover the world’s first known case of an AI-enhanced investigation by a California audit services firm as it uncovered a real human CPA, a controller, committing over $2.8 million in fraudulent transactions. Not only will this change how accounting works, but it may also start restoring confidence in the profession and our financial institutions.
Why does this matter? Fraud has been and will continue to be a massive drain on the world’s economy, with us as citizens bearing the brunt of its effects. According to the Association of Certified Fraud Examiners’ (ACFE) “Report to the Nations” (download required), the potential total global loss caused by fraud is approaching an estimated $4 trillion. There are many AI and machine learning solutions for high-volume, low-amount fraud problems, but while corporate white-collar crimes (financial statement fraud) are the least common cases, they are also the costliest. This means that the bulk of the problem has been left unresolved, and AI is the only realistic and proven technology to help clean this up by doing the impossible: processing, understanding and reporting on never-before-seen insights into our double-entry economy in seconds.
California audit services firm Gursey Schneider LLP used AI-enhanced audit tools to process and understand a massive amount of client data and extract insights with enough evidence to move forward with a $2.8 million criminal fraud case. This type of analysis of over six million transaction records would be impossible with traditional tools such as Microsoft Excel, computer-assisted accounting tools (CAAT) or robotic process automation (RPA), but it falls right within the sweet spot of the capacity and speed of AI. Its potential to ingest vast amounts of information — a must in today’s big data world — and understand the details of what’s going on under the hood far surpasses anything that older methods can accomplish. To date, only a few of the Big Four have had the opportunity to invest in and explore the possibilities of AI, but there are no known reported cases of them having this type of success. With the democratization of AI and surge of venture capital, we’re seeing the advent of a new class of solutions.
Yet this doesn’t mean that accountants, auditors, fraud and forensics experts, or any finance professionals should fear for their jobs. Rather, it’s the opposite. While AI can take 100% of the data and identify the suspect transactions, it still requires human context and intuition and client understanding to pull it all together and report to the client. In fact, Gursey has seen the perceived value of services they deliver reinforced thanks to this engagement.
With today’s challenges in the amount and complexity of data, and increasing scrutiny by regulators, firms must set a road map to AI for themselves. Given that 42% of fraudsters committed their acts by creating fraudulent transactions in accounting systems, it’s the only method capable of detecting schemes. How does AI help? The biggest advantage over traditional methods is that it unlocks anomalies and rare occurrences hidden in the deep mountains of data, where no human could possibly have the time or energy to inspect. AI does not necessarily flag instances of fraud directly — it still takes a human’s intuition and experience to make the final judgment call — but it does identify related patterns of activity across the full data set and calls out:
• Rare and unusual transactions: Flows between accounts that do not match the frequency characteristics of the data set.
• Outliers: Transactions outside the norm, a common method used by fraudsters to cover their tracks within similar, legitimate transactions.
• Expert rules: Transactions outside the learned knowledge base of the system as input by real auditors, also known as “expert systems.”
While AI will never replace the need for finance professionals, those who don’t embrace the technology will be replaced by those who do. So, my recommendation is to look seriously into the technology and learn how it fits within your firm. Let’s change the definition of how fraud investigation really works.