Skip to content

AML automation with AI: From manual screening to intelligent triage

Anti-money laundering (AML) is the prevention of money laundering and terrorist financing. In Spain, Law 10/2010 requires credit and financial institutions to apply customer due diligence and to report suspicious transactions to SEPBLAC; Regulation (EU) 2024/1624 harmonizes it from July 2027.Read more → AML & Artificial Intelligence · May 2026 · 7 min read

AML compliance concentrates most of its cost in repetitive tasks: manual alert review, sanctions list screening, and report writing. AI is changing the rules.

The problem: alert fatigue

Credit institutions estimate, in figures provided confidentially to the European Commission, that between 50% and 75% of the reports received by financial intelligence units are false positives (SWD(2021) 190, July 2021). Most review work therefore goes into discarding noise. The result:

ProblemImpact
Alert fatigueDesensitized analysts who may overlook real risks
Operational costOversized teams to cover manual review volume
Response timeWeeks between alert detection and case resolution
InconsistencyDisparate criteria between analysts for the same alert type

Where AI adds real value

1. Intelligent sanctions screening

Traditional screening compares exact names against lists. The problem: transliteration variations, aliases, and typos drive up the volume of false positives.

AI solution:

  • Fuzzy matching with Dice coefficient to detect spelling variations
  • Contextual analysis weighting jurisdiction, sector, and client history
  • False positive learning that improves accuracy with each review

2. Automatic alert triage

Instead of analysts reviewing each alert from scratch, AI pre-classifies and enriches:

StepManualWith AI
Alert receptionAnalyst opens caseSystem classifies priority (High/Medium/Low)
EnrichmentManual search in 5+ databasesAutomatic 360° KYC profile with 8 risk factors
Preliminary analysisDocument and transaction reviewExecutive summary generated by LLM with key indicators
DecisionAnalyst decides whether to escalateAI proposes action, analyst confirms (four-eyes)

3. Automatic report generation

The special examination under Article 17 of Law 10/2010 requires exhaustive documentation and precedes the indication-based report to SEPBLAC (Servicio Ejecutivo de la Comisión de Prevención del Blanqueo de Capitales e Infracciones Monetarias) is Spain's Financial Intelligence Unit and its supervisor for the prevention of money laundering and terrorist financing. Obliged entities file suspicious transaction reports with it (Law 10/2010).Read more → SEPBLAC (Article 18). AI can:

  • Pre-fill the F19 (form F19-1) is the template an obliged entity uses to report to SEPBLAC any act or transaction with indications or certainty of money laundering or terrorist financing, under Article 18 of Spain's Law 10/2010 and after the special examination of Article 17. Its content and audit trail are reviewed in inspections.Read more → F19 form with case file data
  • Draft the case narrative summary in regulatory language
  • Cite applicable SEPBLAC indicators with regulatory basis
  • Generate the chronological timeline of suspicious operations

Sovereign AI: why processing location matters

In AML, the data processed by AI includes specially protected information: full names, tax IDs, bank details, transaction patterns. Sending this data to third-party APIs poses serious problems:

RiskDescription
Data sovereigntyData leaves your jurisdiction without control
GDPR Art. 28AI provider becomes data processor
AI ActHigh-risk AI systems require transparency (Art. 13) and data governance (Art. 10)
Banking secrecyInformation subject to confidentiality regulations

The alternative: sovereign processing with models running on your own infrastructure (vLLM, Ollama). Data never leaves the controlled perimeter.


How BlueUPALM implements AML automation

CapabilityImplementation
ScreeningReal-time fuzzy matching against EU, OFAC, UN lists
TriagePre-classification with 9 configurable SEPBLAC indicators
Four-EyesSegregation of duties with Four-Eyes principle
ReportingAutomatic F19/CXI generation with cryptographic audit trail
Sovereign AILocal processing with vLLM — data never leaves the perimeter

Want to see AML automation in action?

We'll show you the BlueUPALM AML engine with synthetic data from your sector.

Request AML demo


Back to blog · View BlueUPALM product