AML (Anti-Money Laundering)

Anti-Money Laundering (AML) is the body of laws, regulations, and institutional controls aimed at preventing criminals from disguising illicit funds as legitimate money. Financial institutions are legally obligated to know who their customers are, understand the source of their funds, monitor transactions for suspicious patterns, screen against sanctions and watchlists, and report suspicions to authorities. Regimes differ by jurisdiction — the Bank Secrecy Act in the US, EU AML directives, FATF recommendations globally — but the operational shape is similar everywhere, and the penalties for failure run to billions.

AML is, operationally, a document-intensive discipline. Customer due diligence runs on identity documents, proof of address, corporate registries, and ownership structures; source-of-funds and source-of-wealth checks run on bank statements, contracts, tax filings, and sale deeds; enhanced due diligence adds adverse media and complex corporate documentation. Every one of those documents must be read, verified, extracted, and reconciled — which is why AML operations historically meant floors of analysts, and why document AI has become central to modernizing them: automated extraction from ID and financial documents, forgery and tampering detection, entity resolution across documents, and screening pipelines that read the news the way analysts do.

The compliance constraint shapes the technology choices. AML decisions must be explainable and evidenced — a regulator will ask why an alert was closed, what documents supported an onboarding decision, and how the institution knows its controls work. That favors document AI with per-field provenance, confidence-based human review, and audit trails over black-box automation. And because the underlying documents are among the most sensitive an institution holds, processing them within the institution's own perimeter — rather than through external APIs — is increasingly treated as part of the control environment itself.