Client Alert: Latest Guidance on the Use of AI for FCC

Overview

The Wolfsberg Group recently issued the “Wolfsberg Principles for Using Artificial Intelligence and Machine Learning in Financial Crime Compliance”. This announcement outlines key considerations for financial institutions (FIs) currently using, or considering using, artificial intelligence technology to support their financial crimes compliance (FCC) program.  The Principles consist of five elements that support an FI’s responsible use of AI/ML in financial crime compliance applications:

• Legitimate Purpose
• Proportionate Use
• Openness and Transparency
• Design and Technical Expertise
• Accountability and Oversight

Three of the Principles are designed to align an FI’s use of AI technology, and the data used to support it, with the data ethics principles founding significant data privacy regulations, such as the EU’s General Data Protection Regulation (GDPR).  Legitimacy of purpose, proportionate data use and transparency are principles that each FI needs to define for itself as it contemplates the use of AI technology to support its FCC program.

The remaining two Principles are founded on more technical considerations and represent industry best practice for the responsible development and use of AI technology.  FIs considering or using vendor-provided AI solutions should consider the Wolfsberg Principles on AI Design, Testing and Oversight to ensure that their vendor’s AI solution meets the specific requirements of these principles:

• AI Design and Testing:  AI based compliance systems should be:
  • Created, monitored, and controlled by staff with the appropriate skills and diverse experiences needed to identify bias in the results
  • Developed with a clear definition of the intended outcomes
  • Capable of generating explainable results given the data inputs
  • Subjected to ongoing testing, validation, and re-configuration to review AI outcomes based on the intended purpose
• Appropriate Oversight:  Staff with access to the inputs/outputs of AI based compliance systems should be trained on the appropriate use of data inputs, AI technology, and the handling of results. FIs should also establish processes to challenge their technical teams whenever necessary and probe the use of data within their organizations.

FIs should expect vendors providing them with AI technology to be able to provide them with the documentation required to support the alignment of their AI-enabled compliance solutions with the key elements of AI development, testing and oversight, which at a minimum should include documentation of the AI development and ongoing oversight processes:

• How the AI technology has been created with a clear definition of the intended outcomes
  • That the AI was created and is actively monitored and controlled by vendor staff with the appropriate skills and diverse experiences needed to identify bias in the results
  • How the AI technology generates explainable results given the data inputs,
  • The processes used for the ongoing testing, validation, and re-configuration of AI technology to conform AI outcomes to the intended purpose
  • Training provided by the vendor to FI staff with access to the inputs/outputs of AI-based compliance systems on the appropriate use of data inputs, AI technology and the handling of results

How Exiger Can Help

Exiger provides a diverse range of FIs, Corporates and Government Agencies with AI-enabled compliance technology solutions for customer name screening, transaction screening, vendor management and supply chain risk management.  All of Exiger’s AI-enabled compliance solutions were designed and continue to operate in full alignment with these principles of AI development, testing and oversight. We routinely provide our clients with the documentation needed to support their alignment with them. Exiger can also evaluate homegrown AI models for alignment with the Wolfsberg AI principles.

Contact us today to discuss how Exiger can support your program.