Role of artificial intelligence and machine learning in detecting and combating financial crimes

The resulting transactions, amounting to millions of Nigerian Naira, had passed through numerous accounts before vanishing without a trace. He storms into the nearest branch of his bank, bewildered and angered, the branch manager could only offer a tired sigh as it was not an isolated case. Somehow, somewhere, an unseen network of criminals were exploiting the bank’s internal system, moving customers’ funds at an alarming speed, faster than any human could track.

In the evolving landscape of financial technology where digital transactions dominate global economies, financial crimes are no longer a backroom operation carried out by a handful of fraudsters; it has morphed into a highly sophisticated global industry, costing economies trillions of US dollars annually. From cybercriminals laundering illicit funds through complex networks to scammers and fraudsters exploiting vulnerabilities in the digital banking ecosystem, financial crime has become a formidable challenge to financial institutions and their regulators alike.

Global financial crime has reached crisis levels, with far-reaching consequences. According to a report by the United Nations Office on Drugs and Crime (UNODC), money laundering figures alone are about US$2 trillion annually, equivalent to nearly 5% of global GDP. In the United States, the Federal Trade Commission (FTC) reported over US$8.8 billion in various financial and investment fraud losses in 2022, marking a 30% increase from the previous year. Similarly, the Nigerian Inter-Bank Settlement System (NIBSS) reported that fraud-related losses in Nigeria surged to ₦9.5 billion (US$12.3 million) in 2023, highlighting the urgent need for advanced detection mechanisms.

We have moved beyond the times when analysts’ primary reliance were on intuition, manual processes, rule-based monitoring, and reactive compliance to uncover fraudulent activities. These traditional processes are no longer a match for the speed and complexity of modern-day financial crimes. In recent times and with the emergence of technological innovations that birthed artificial intelligence and machine learning, these tech innovations have revolutionised the way financial institutions identify and mitigate financial crimes.

According to a 2021 study by Deloitte highlighting artificial intelligence’s potential, “AI can analyse complex transactions, identify potentially fraudulent activities, and even predict future fraudulent transactions with remarkable accuracy.” This predictive capability of artificial intelligence is particularly valuable in the financial sector where staying one step ahead of criminals is crucial to prevent financial crimes.

With the help of artificial intelligence and machine learning, financial institutions can now analyse vast amounts of data sets in real time, uncovering patterns and anomalies that may indicate financial crime activities. Moreso, it is pertinent to state that artificial intelligence doesn’t just detect financial crimes; it also predicts financial crimes, learning from past patterns to identify anomalies prior to the resulting devastating losses.

One of the most significant advantages of artificial intelligence and machine learning in detecting and combating financial crimes is their prowess in recognition. This is because financial crimes often involve intricate networks of transactions designed to obfuscate illegal activities.

A report by the International Data Corporation (IDC) highlights that global spending on artificial intelligence in the financial sector is expected to reach $11 billion by 2025, reflecting the growing reliance on these technologies to combat financial crimes. This investment is driven by the ability of artificial intelligence and machine learning to enhance the speed and accuracy of detecting financial crimes.

In the realm of anti-money laundering (AML) and counter-terrorist financing (CTF), the deployment of artificial intelligence helps scour transaction networks to identify hidden patterns that indicate illicit activities. Money laundering typically involves disguising large sums of ill-gotten funds into smaller transactions that appear legitimate in a process known as ‘smurfing’. However, the deployment of artificial intelligence can connect these dots in ways humans cannot, tracking money as it moves through multiple accounts and detecting suspicious behaviours. A study by McKinsey found that AI-powered AML tools improved detection rates by 50% while reducing compliance costs by 30%, offering financial institutions a significant advantage.

In addition, market manipulation and insider trading are other areas where AI is proving invaluable. Globally, stock exchanges handle millions of trades daily, making it nearly impossible for human regulators to monitor every transaction for foul play. Meanwhile, with the help of AI-driven surveillance systems, it is easier to analyse trading patterns, news reports, and even social media sentiment to flag irregular activity.

Beyond direct crime detection, artificial intelligence plays a crucial role in risk assessment and regulatory compliance. Financial institutions are required to screen clients and transactions against global watchlists, sanction lists, adverse media hits and politically exposed persons (PEP) databases. Traditionally, these compliance processes take longer time. However, in recent times, AI-driven solutions can complete due diligence within minutes, drastically improving efficiency. According to IBM, AI-based compliance tools have reduced manual workloads by 80%, allowing banks to focus on high-risk cases more effectively.

Despite its strengths, AI is not a standalone solution as it is not without its challenges. Human expertise remains indispensable to achieve optimal functionality of any tech solution. Artificial intelligence-powered solutions can highlight potential threats, but human analysts must assess whether an alert is genuinely suspicious or just an unusual but legitimate transaction. A study by PwC found that 67% of financial crime professionals believe AI enhances rather than replaces human intelligence in fraud detection, ensuring that ethical oversight and contextual judgment remain part of the process.

Moreover, one major concern of artificial intelligence is bias. If an AI model is trained on skewed data, it might disproportionately flag transactions from certain demographics, leading to unfair scrutiny. There are also ethical concerns surrounding privacy because AI-driven fraud detection involves monitoring vast amounts of personal financial data, raising questions about data protection and regulatory compliance under laws like the General Data Protection Regulation (GDPR). Financial institutions must navigate these complexities carefully, ensuring that AI is used responsibly.

Criminals, too, are adapting. Just as AI strengthens security, fraudsters are using the same technology to create more convincing scams. Deepfake technology is being used to generate realistic fake identities, and AI-generated phishing attacks are becoming harder to distinguish from genuine communications. This cat-and-mouse game means financial institutions must constantly update their AI models to stay ahead of evolving threats.

The future of financial crime detection will be shaped by AI’s continued evolution. With AI-driven fraud prevention reducing false positives, improving compliance, and detecting threats in real-time, financial institutions are better equipped to protect themselves and their customers. However, the key to success lies in the balance of leveraging AI’s power while maintaining human oversight and ethical responsibility.

As financial crimes grow more complex, so too must the tools used to combat them. The fight against financial crime is no longer just a battle of rules and regulations; but rather a war of precision, intelligence, speed, and technology.

The article is written by Olufemi Dada, a financial crime system analyst.