Artificial Intelligence is being used to automate tasks in manufacturing, transportation and data analysis, processing vast quantities of data at rapid speeds with the click of a button. As AI becomes more affordable and accessible, many firms are starting to use AI to improve productivity and simplify basic business processes. AI has evolved and become accepted by mainstream organisations to such a degree that leading financial institutions like banks and fintech businesses are starting to use AI in developing and implementing their anti-money laundering processes. 

This does not mean that using AI in AML is without risk, however. The Wolfsberg Group, a global association of leading banks, has developed a set of guidelines to help financial institutions use AI in their anti-money laundering frameworks in an effective and responsible manner. 

What Are Some of the Concerns About Using AI for AML?

There has been some criticism regarding the use of AI in anti-money laundering initiatives. Unlike conventional software tools, AI systems are not programmed with specific rules or instructions but instead are provided with a set of examples (training data) to learn to perform tasks, a process known as machine learning. Some critics believe that AI systems may perpetuate or exacerbate existing biases if not properly trained against data that is representative of the population, which in turn impacts accuracy. After all, AI systems are only as good as the data they are trained on. Poor quality data will result in inaccurate results and poor performance. This can be particularly challenging as sanctions and PEP data (as well as AML regulations) continually change. 

Many human operators may also have difficulty understanding and interpreting AI systems, which can hamper necessary interventions and even the machine learning process itself. There is also a question of whether or not financial institutions will become overly reliant on AI systems, neglecting other relevant controls and missing important money laundering activity as a result. AI systems, especially when designed by companies with a limited understanding of regulatory requirements, may make decisions that are not in the best interest of the financial institution. 

Others have raised privacy and cybersecurity concerns as AI processes large amounts of sensitive data. If the AI system isn’t properly secured, it may compromise the integrity of the data. Some are concerned that AI will be misused for money laundering and illegal activities if not managed and controlled properly. 

The Wolfsberg Group Principles outline appropriate measures to mitigate these risks, including the implementation of effective governance, data protection, and security measures and ensuring that there is sufficient, expert human oversight of AI systems. 

What Are The Wolfsberg Principles? 

The Wolfsberg Group is an association of 13 global banks that joined forces in 2000 to promote and develop financial transparency, accountability and integrity. Members include some of the world’s most prominent banks, including JPMorgan Chase, CitiGroup, and Deutsche Bank. The collaboration is best known for developing the Wolfsberg AML principles, a set of guidelines for financial institutions to use in their fight against money laundering and terrorist financing. 

Other guidelines published by the group details various best practices, including the Wolfsberg Due Diligence Guidelines for Correspondent Banking, the Wolfsberg Statement on the Provision of Financial Services to Politically Exposed Persons (PEPs) and The Wolfsberg Principles for the Use of Artificial Intelligence in Anti-Money Laundering (AML).

While the Wolfsberg Group is a private organization without regulatory powers, its guidelines have been adopted by financial institutions around the world and are often seen as a benchmark for compliance. 

The Wolfsberg Principles for the Use of Artificial Intelligence were designed to guide financial institutions, particularly their financial crime compliance leaders and risk managers, in identifying operational risks that may arise from the use of artificial intelligence. 

The Principles consist of five elements that outline the responsible use of AI and machine learning in financial crime compliance, including: 

  • Having a legitimate purpose; 
  • Proportionate usage; 
  • Technical expertise in design and application; 
  • Accountability and oversight; and 
  • Openness and transparency. 

The Wolfsberg Principles in Practice

The Wolfsberg Principles for the Use of AI in Anti-Money Laundering covers several critical areas surrounding the use of artificial intelligence in compliance efforts. 

The Group recommends that financial institutions establish a robust governance framework that can evaluate the use of AI in AML to ensure that it is consistent with legal, regulatory and internal requirements. Financial institutions must be transparent with both their customers and regulators about the way AI is used in their AML efforts, including the data, algorithms and decision-making processes used. AI systems must be robust and explainable, e.g. the system should function correctly and provide clear reasoning for its decisions. AI should not discriminate against any individuals or groups or perpetuate existing biases. 

Data used to train and evaluate AI systems should be of high quality and protected against unauthorised access and use. AI must be subjected to appropriate, rigorous human oversight and teams responsible for overseeing the system must have the ability to intervene and override automated decision-making processes. 

It’s important for firms to note that having an AI system in place does not excuse them from their regulatory and legal responsibilities - they must be able to demonstrate that their system can meet their compliance requirements. AI systems should be continuously monitored and improved to ensure that the firm remains fully compliant at all times. 

Final Thoughts

AI can be an extremely useful tool for financial institutions that have to process and verify a large volume of data and transactions, but AI is only as good as the provider that trains and manages the system itself. When choosing an AI-based screening solution for AML, financial institutions should take care to ensure that the company has the right level of experience and expertise in AI.

If you would like to know about how to effectively use and implement an AI-based screening solution, get in touch with our customer support team.