Wolfsberg Group Urges Financial Institutions to Adopt AI for Monitoring Suspicious Activities

The Wolfsberg Group, in its latest statement, has urged banks to accelerate the use of artificial intelligence and data-driven systems to monitor suspicious activity, replacing outdated tools that generate large numbers of false alerts while failing to capture the highest-risk activity.

The guidance, published Sept. 9 as Part II of the Wolfsberg Group’s Statement on Effectiveness through Monitoring for Suspicious Activity, outlines a framework for financial institutions to transition from legacy rules-based systems to innovative models using machine learning and automation in order to detect suspicious activities and prevent financial fraud.

The 13-member association of global banks said the changes are needed as criminals evolve faster than the rules and traditional compliance. The group mentioned a statement by the US Supervisory agencies encouraging innovation in financial crime models to match the pace of new threats. 

They highlighted that the US regulatory bodies, 

“Support efforts by banks to innovate and update [their financial crime] systems and models to quickly adapt to an evolving threat environment.”

Banks are encouraged by the group to stop comparing new monitoring systems with outdated ones and instead evaluate them using modern performance measures, such as accuracy, how effectively suspicious activity is detected, whether key risks are covered, and how useful suspicious activity reports (SARs) are to law enforcement. It emphasized that instead of relying on old practices like parallel runs, banks should focus on proof of concept testing and historical data evaluation.

The banking association said imperfect models should not be rejected if they improve crime detection, cautioning against treating anti-money laundering (AML) tools with the same governance standards as credit or market risk models.

The guidance further states that financial institutions must be transparent about how models work, how they are calibrated, and how investigators should interpret alerts. Hybrid approaches combining rules-based systems with supervised and unsupervised machine learning are recommended.

The Wolfsberg Group urged banks to cut duplication in audit and oversight, warning that overlapping checks slow down new monitoring methods. Instead, it called for governance to match the real importance of each model, based on factors like product coverage, transaction volume, and its impact on crime detection.

The Wolfsberg Group first addressed effectiveness in monitoring suspicious activity in July 2024 with Part I of its statement, which urged banks to measure programmes by their value to law enforcement rather than volume of alerts. Part II expands on that by providing detailed guidance for innovation and transition.

The association of banks further points out that effective monitoring requires banks to accept changes in risk appetite, adapt internal oversight, and train investigators to handle AI-driven cases. The statement Innovation must be explainable, risk-based, and aligned with national priorities.

The Group’s members include global banks such as JPMorgan Chase, HSBC, Citigroup, and UBS. The Wolfsberg framework is built on three principles: transitioning and validating new systems, balancing model risk against financial crime risk, and ensuring transparency. Its principles are widely referenced by regulators and financial institutions when shaping anti-money laundering.