Based on a global survey of 850 members of the Association of Certified Anti-Money Laundering Specialists (ACAMS), the study reveals:

• AI and machine learning (ML) adoption remains modest. Only 18% of respondents report having AI/ML solutions in production. Another 18% are testing AI/ML solutions, while 25% plan to deploy AI/ML in the next 12-18 months; 40% have no current plans to adopt AI/ML.
• Interest in generative AI technology is strong but apparently cautious. Nearly half of respondents say they currently have generative AI systems in testing (10%) or are in the discovery phase (35%) - a significant percentage for an emerging technology. However, 55% have no plans to adopt generative AI.
• The Road to Integration: State of AI and ML Adoption in AML Compliance, a follow-up study to a similar survey published in 2021, explores the current state of AI/ML adoption to combat money laundering. SAS also published a data dashboard that allows users to explore, visualize and filter the survey information by region and institution size.

"Key to unlocking the full potential of AI and ML is the integration of data sources, teams and technology. The first step toward this integration is establishing a data ecosystem that combines information from all sources," said Stu Bradley, Senior Vice President at SAS. "Some organizations may be waiting for guidance from regulators. Companies that move forward with data and operations integration with governance in mind are laying the groundwork for responsible innovation in artificial intelligence and machine learning and will have a competitive advantage over those that hesitate."

AI and ML produce value when fully deployed

The survey provided more insights into how AI technology is being used to combat money laundering and why companies are slow to fully integrate it:

• Organizations are identifying multiple uses for AI/ML. In the 2021 survey, 78% of respondents cited either improving the quality of investigations and regulatory findings (40%) or reducing false alarms (38%) as the top reason for adopting AI/ML. This year, these two responses remain the top two, but their combined percentage dropped 11 points to 67%. Complex risk detection increased from 17% to 21%, and "None of the above" jumped from 5% to 13%.
• Reducing false alarms is becoming a growing priority. When asked about the priorities for AI/ML, AML experts cited reducing false alarms in existing surveillance systems to 38% (an 8% increase from 2021).
• Machine learning has a significant impact - but NLPs shouldn't be neglected either. When asked to rank three technologies by impact, machine learning was again the top choice at 58%, up 6% from 2021. Robotic process automation (RPA) fell to 28%, while natural language processing (NLP) was last at 14%.