In the aftermath of South Africa’s recent greylisting by the Financial Action Task Force (FATF), companies are now confronted with the imperative to address eight identified strategic deficiencies while simultaneously reducing their financial crime risk through anti-money laundering (AML) compliance processes.
These challenges, notorious for their potential cost and time commitments, underscore the complexity of achieving full compliance. It’s little wonder that a staggering two-thirds of businesses are charting a course towards heightened technology investments to wage war against financial crime and reinforce compliance.
According to James Saunders, chief technology officer and co-founder of RelyComply, artificial intelligence (AI) is swiftly emerging as a potent tool to revolutionise antiquated, manual risk-mitigation methods. “This technology brings precise data-driven insights to the table that are markedly less susceptible to human error, although human involvement remains irreplaceable for success.”
Elevating AML through AI
He explains that AI serves a dual purpose within AML compliance protocols: task automation and advanced data analysis, with the first application centring on streamlining time-consuming tasks.
“For instance, AML analysts can harness automation to summarise documents, gauge messaging sentiment, or extract significant adverse media. It also aids in verifying customer information and tracking interactions over time. This approach can often be more cost-effective than relying on off-the-shelf tools which may or may not outperform their human counterparts.”
As for the second application, Saunders underscores AI’s prowess in processing vast datasets to discern patterns and flag anomalies. “No human can rival a computer’s innate data processing capabilities, which prove invaluable for transaction monitoring and synergy with a company’s customised datasets. By assigning roles based on what AI can automate and where human intervention is necessary, AML and CFT processes could be significantly streamlined.”
Understanding AI risk
He acknowledges that AI’s evolution comes with its own set of challenges. “Evaluating how AI, at various maturity levels, fits into existing business processes is no straightforward task. This is because traditional AML processes often rely on rules-based systems that can miss errors or trigger ‘risky’ data flags that aren’t substantiated (resulting in false positives).
“A staggering 95% of system-generated alerts are reported as ‘false positives’, potentially leading to stringent regulatory actions, real-world repercussions, and damage to professional reputations.
“While AI has seen initial use in low-risk cases, its application for AML compliance is far from simplistic,” adds Saunders. “Misguided use of AI in AML compliance could, at worst, raise concerns about customer understanding and erode trust in financial systems. For example, facial recognition technology has exhibited biases in race and gender identification. AI algorithmic bias could perpetuate this by generating incorrect risk profiles for customers based on unrelated fraudulent activities within their jurisdiction.
“Though AI offers significant advantages in AML compliance, firms must possess a thorough understanding of its potential, necessitating input from AI specialists to make informed decisions about its implementation,” he highlights.
AI and humans: a synergistic approach
When it comes to the collaboration between AI and humans to ensure more dependable AML compliance, Saunders observes that AI can surpass the confines of a single metric. “It can identify the context and characteristics of specific transactions, potentially predicting future criminal activity based on learned patterns.
“By effectively tagging and indexing customer or transaction data, AI empowers humans to visualise crucial information that might otherwise be overlooked while minimising the risk of false positives.
“AI-backed technology accelerates data processing, enhances risk comprehension, and provides immediate audit evidence. However, it ultimately falls to a compliance officer, armed with the correct financial data, to make critical decisions,” he notes. “Both humans and AI tools strive to achieve the same objectives, each reinforcing the other.
“AI, on its own, may not be a silver bullet. But a revamped compliance solution driven by automation and advanced data analysis could be a game-changer in the battle against money laundering when coupled with experts in the field,” concludes Saunders.