Skip to main content

Looking for Valuant? You are in the right place!

Valuant is now Abrigo, giving you a single source to Manage Risk and Drive Growth

Make yourself at home – we hope you enjoy your new web experience.

Looking for DiCOM? You are in the right place!

DiCOM Software is now part of Abrigo, giving you a single source to Manage Risk and Drive Growth. Make yourself at home – we hope you enjoy your new web experience.

Looking for TPG Software? You are in the right place!

TPG Software is now part of Abrigo. You can continue to count on the world-class Investment Accounting software and services you’ve come to expect, plus all that Abrigo has to offer.

Make yourself at home – we hope you enjoy being part of our community.

AML/CFT AML Software Financial Crime Fraud Prevention

Human-in-the-loop: Enhancing AI in AML/CFT software for competent case management

Terri Luttrell, CAMS-Audit, CFCS
September 24, 2024
Read Time: 0 min

When AI complements human decision-making

Human oversight, often called a human-in-the-loop approach, remains important as AI increasingly enhances AML/CFT software. 

You might also like this checklist, "6 steps for compliance with the new AML/CFT program rules."

DOWNLOAD

 

Takeaway 1

A human-in-the-loop approach plays a vital role in ensuring that AI systems effectively support alert and case management for AML/CFT suspicious activity monitoring.

Takeaway 2

In an AI-driven AML/CFT workflow, human oversight allows institutions to adjust algorithms, fine-tune results, and handle the more complicated cases that AI might misinterpret.

Takeaway 3

Financial institutions evaluating AML/CFT software have several compliance considerations related to involving humans in processes. 

AML/CFT efforts

Human oversight remains crucial

As financial institutions increasingly adopt artificial intelligence (AI) to enhance anti-money laundering and counter-terrorist financing (AML/CFT) efforts, the need for human oversight remains crucial. This approach, known as human-in-the-loop (HITL), plays a vital role in ensuring that AI systems effectively support alert and AML/CFT case management for suspicious activity monitoring, allowing for human interaction and critical thinking.

The Financial Crimes Enforcement Network (FinCEN) recently released proposed legislation that encourages innovation within AML/CFT programs, advocating for the integration of advanced technologies while maintaining compliance through human supervision​. The proposed AML/CFT program rule encourages innovation to streamline compliance with AML/CFT obligations, expecting financial institutions to consider, evaluate, and, when appropriate, implement innovative technologies such as AI and machine learning. However, AI alone does not create a robust AML/CFT program or satisfy regulatory expectations. Investigative expertise is critical.

Collaboration between AI and people

What is human-in-the-loop?

Human-in-the-loop refers to the collaboration between AI systems and human experts, ensuring that machines do not operate in isolation. While AI can process vast amounts of data quickly, detect patterns, and even reduce false positives in suspicious activity monitoring, it lacks the contextual understanding and decision-making capabilities that experienced AML professionals provide. As AML/CFT investigators understand, the BSA world is not black and white. While some alerts are simple and can easily be handled by AI, many are complex and require extensive analysis. 

In an AI-driven AML/CFT workflow, human oversight allows financial institutions to adjust algorithms, fine-tune results, and handle the more complicated cases that AI might misinterpret. Suspicious activity in an AML/CFT program cannot be missed.

Learn how Signature Bank of Georgia enhanced fraud detection. READ STORY

Context, complexity, adaptation

Why is HITL important in AML/CFT?

The approach that keeps a human in the loop is essential in alert and case management for several reasons:

  1. Contextual judgment: AI excels at spotting unusual patterns, but not all anomalies indicate fraud or money laundering. Human experts use contextual judgment to differentiate between normal behavior and illicit activities. AI reduces the risk of costly false positives for simple alerts that can drain resources and leave the complexities to human analysis.
  2. Handling complex cases: Certain AML/CFT cases involve complex data layers or require an understanding of industry-specific regulations. A purely AI-driven approach may miss these nuances. Human investigators ensure the broader picture is considered, especially in high-risk cases that demand more than just data-based conclusions.
  3. Adapting to new threats: Financial crime evolves quickly, with fraudsters continuously finding new ways to exploit systems. AI models need regular updates and supervision to adapt to these new threats. By integrating human expertise, institutions can ensure AI is continually learning from emerging threats and regulatory changes, preventing outdated models from leaving blind spots and improving risk management.
  4. Regulatory compliance: AML/CFT regulations require institutions to justify their actions and decisions during audits or examinations. HITL ensures that AI recommendations are backed by human review and documentation, providing an additional layer of validation that satisfies regulatory requirements.
  5. Building trust in AI: Financial institutions that are new to AI may hesitate to rely entirely on machines for critical tasks like AML/CFT compliance. Human oversight builds confidence in AI systems by assuring that machine-generated results align with professional standards and regulatory expectations. Testing of any model is critical to the success of any AML/CFT program.

Examples of human intervention

Human-in-the-loop in action

Consider an example where an AI system flags a series of complex transactions as potentially suspicious transactions. Without human intervention, these transactions may automatically escalate to a case, consuming valuable investigative resources. With HITL, an analyst can review the flagged activity, assess its risk, and determine whether it merits further investigation. In many instances, human experts can identify patterns or customer behavior that the AI might not fully grasp, preventing unnecessary case escalations and reducing the compliance team’s workload.

On the flip side, AI can detect typical patterns from alerts that may not be suspicious, such as reviewing potentially fraudulent checks within fraud detection software and not sending straightforward alerts to cases. This results in a reduction in false positives, leaving human investigative resources for the more complex, higher-risk cases.

Compliance considerations

Evaluating HITL in AML/CFT software

When evaluating an AML/CFT monitoring solution, financial institutions should carefully consider how much HITL is necessary for their compliance program. Here are some key considerations:

  • Institutions should assess their risk-based AML/CFT program to determine what degree of HITL is needed to avoid regulatory criticism. If the institution operates in high-risk markets, such as money services businesses or cannabis, or deals with complex customer profiles, human involvement can provide the additional scrutiny needed to ensure that suspicious activity is not missed.
  • Once the level of human intervention is determined based on the institution’s risk profile, ask the right questions during AML software due diligence:
    • How does the system incorporate human oversight into AI decision-making?
    • Does the case management feature allow for humans to intervene in high-risk cases or flagged alerts?
    • What roles do AML/CFT experts play in adjusting AI models and improving accuracy?
    • How easy is it to review, override, or update AI-generated recommendations?

Conclusion

AI is a powerful tool and is very useful in streamlining AML/CFT suspicious activity monitoring, but its effectiveness is strengthened with the involvement of human experts. HITL ensures that technology complements human decision-making rather than replacing it. For financial institutions, this collaborative approach leads to program confidence, more accurate alert management, better resource allocation, and more robust compliance outcomes. By balancing the strengths of AI and human expertise, institutions can enhance their fight against financial crime while maintaining the confidence and soundness of their AML/CFT program.

This blog was written with the assistance of ChatGPT, an AI large language model, and was reviewed and revised by the subject-matter expert.

Don't settle for out-of-the-box AML software if you want the best. BAM+ was built for your unique risk profile.

learn more about bam+
Webinar

Alert & ready: Optimizing BSA/AML & fraud detection tuning for compliance

Read More
Webinar

Exposing pig butchering fraud: A case study from HSI

Read More
Webinar

AML/CFT Requirements: Are you ready for FinCEN’s final rule?

Read More
About the Author

Terri Luttrell, CAMS-Audit, CFCS

Compliance and Engagement Director
Terri Luttrell is a seasoned AML professional and former director and AML/OFAC officer with over 20 years in the banking industry, working both in medium and large community and commercial banks ranging from $2 billion to $330 billion in asset size.

Full Bio

About Abrigo

Abrigo enables U.S. financial institutions to support their communities through technology that fights financial crime, grows loans and deposits, and optimizes risk. Abrigo's platform centralizes the institution's data, creates a digital user experience, ensures compliance, and delivers efficiency for scale and profitable growth.

Make Big Things Happen.