FORMATION IA
IA pour les équipes criminalité financière, LCB-FT et KYC
Outillez vos équipes conformité et fraude pour évaluer, déployer et gouverner l'IA dans la détection de la criminalité financière.
Ce qu'elle couvre
Ce programme de niveau praticien couvre l'application de l'IA et du machine learning dans la lutte contre le blanchiment d'argent, les processus KYC et la détection des fraudes. Les participants apprennent le fonctionnement des systèmes modernes de surveillance des transactions, l'utilisation de la résolution d'entités et de l'analytique de graphes pour détecter les réseaux cachés, et comment évaluer les offres des fournisseurs au regard des attentes réglementaires (ABE, FinCEN, GAFI). Le programme combine études de cas, exercices pratiques d'évaluation de fournisseurs et un atelier d'amélioration de la qualité des déclarations de soupçon.
À l'issue, vous saurez
- Critically evaluate an AI transaction monitoring vendor's model card against EBA and FATF explainability expectations
- Map a money laundering typology to the appropriate graph analytic or ML detection technique
- Design a SAR quality improvement workflow using AI-assisted narrative generation with human review gates
- Build a vendor RFP scorecard that stress-tests false-positive rates, model drift policies, and audit trail requirements
- Articulate the model risk management obligations (SR 11-7 / SS1/23) applicable to an AI fraud model in production
Sujets abordés
- How ML-based transaction monitoring differs from rule-based systems and when each is appropriate
- Entity resolution and graph analytics for uncovering hidden beneficial ownership and money mule networks
- AI-assisted SAR drafting: improving narrative quality, reducing false positives, and meeting regulatory expectations
- KYC automation: document verification, biometric checks, and continuous monitoring with AI
- Regulatory expectations on explainable AI from EBA, FATF, FCA, and FinCEN model risk guidance
- Vendor landscape assessment: evaluating NICE Actimize, Quantexa, Feedzai, Napier, and open-source alternatives
- Model risk management (SR 11-7 / SS1/23) applied to AML and fraud models
- Bias, fairness, and adverse-action obligations in automated financial crime decisions
Modalité
Delivered as a blended programme over three to four weeks: two live instructor-led full-day sessions (remote or on-site) bookend four facilitated 90-minute online workshops. Roughly 60% of time is hands-on — vendor evaluation exercises, anonymised case datasets, and a capstone SAR workflow redesign. Participants receive a typology reference pack, a vendor assessment template, and access to a private cohort Slack channel for peer discussion.
Ce qui fait que ça marche
- Involve compliance, IT, and internal audit jointly from day one so model governance responsibilities are clear before go-live
- Run a structured parallel-run period (at least 90 days) comparing AI alerts against legacy rule outputs before decommissioning old logic
- Establish a documented false-positive review loop that feeds back into model retraining cadences and is visible to regulators
- Secure explicit regulatory engagement (pre-application meetings with FCA, DNB, or BaFin) when deploying novel AI in high-risk detection scenarios
Erreurs fréquentes
- Deploying AI transaction monitoring as a direct rule-engine replacement without establishing a parallel-run validation period, leading to regulatory model risk findings
- Treating explainability as a checkbox rather than designing audit-ready reason codes from the outset, creating SAR defensibility problems
- Underestimating data quality issues — incomplete beneficial ownership data and inconsistent name formats degrade entity resolution accuracy significantly
- Selecting vendors purely on false-negative reduction without negotiating access to model documentation needed for internal validation under SR 11-7
Quand NE PAS suivre cette formation
This programme is not appropriate for a team that has not yet implemented a baseline AML transaction monitoring system — foundational AML process design and regulatory literacy must come first before AI augmentation adds value.
Fournisseurs à considérer
Sources
Cette formation fait partie d'un catalogue Data & IA construit pour les leaders sérieux sur l'exécution. Lancez le diagnostic gratuit pour voir quelles formations sont prioritaires pour votre équipe.