Where AI Delivers Measurable Value in AML

A recent online webinar convened speakers from NICE Actimize and NatWest to examine where artificial intelligence is producing measurable benefits across the anti-money laundering lifecycle. The discussion focused on moving AI projects from pilots into production and on governance needs.

The panel included Nicola Eschenburg, EMEA AML Strategy Leader at NICE Actimize, and Ashley Bostel, Head of Economic Crime Analytics Strategy at NatWest Group. Teresa Connors moderated the session.

Panelists described specific use cases already in operation: automated adverse media research, more precise customer risk scoring in KYC, optimisation of screening rules to reduce false positives, alert triage to prioritise investigator work, automation of routine investigation tasks, and assistance in preparing suspicious activity reports.

Panelists described how machine learning models, large language models and early-stage agentic AI are being used to connect risk signals stored in different systems and teams. By combining transaction histories, customer profiles, public media and screening outputs, these tools can produce a more complete view of customer risk and speed decision timelines.

Institutions reported measurable outcomes following deployment: faster onboarding, fewer unnecessary alerts and more efficient case handling. Panelists noted that results depend on data quality and on how well models are integrated into existing workflows.

On implementation, speakers outlined three approaches: build in-house models, buy third-party solutions, or apply a hybrid mix. The appropriate choice depends on an institution’s data maturity, regulatory expectations and capacity to govern models.

Governance requirements highlighted included model explainability, audit trails, testing regimes and documentation for supervisory review. Panelists emphasised the need for validation and record-keeping to manage operational risk as AI is embedded in compliance processes.

Speakers addressed agentic AI — systems that can carry out multi-step tasks with limited instruction — as a way to automate parts of investigations and routine reporting. They said such tools can speed repetitive work while leaving judgment and final decisions to human investigators, and that regulatory and transparency requirements limit how much decision-making can be delegated.

Panelists recommended that institutions identify operational use cases with measurable gains and establish governance structures that allow controlled scaling of AI within AML operations.