AI-Enhanced Sanctions Screening A risk-controlled predictive compliance Framework

Authors: Pratik chawande

DOI: https://doi.org/10.37082/IJIRMPS.v14.i1.232939

Short DOI: https://doi.org/hbphz8

Country: United States

Full-text Research PDF File:   View   |   Download

Abstract: The current global financial system is under strain due to the growing rate and complexity of increasing sanctions regimes, where the screening systems of the past, based on rules, produce false positives in unsustainable levels of over 95%. Such operational overload burns compliance resources, causes alert fatigue and blurs indications of real danger caused by advanced evasion strategy. It is in this paper that I suggest a new, risk-managed framework in which artificial intelligence (AI) and machine learning (ML) are merged to transform the sanctions compliance paradigm into a reactivity pattern-matching system into a proactive, predictive approach to risk. The multi-layered architecture of the framework takes advantage of advanced natural language processing (NLP) to resolve entities, network analytics to map relationships and a collection of supervised and unsupervised machine learning models to contextually score risk. Most importantly, it includes explainable AI (XAI) to enable transparency in decisions, a well-developed model risk management layer of governance, and a non-alterable audit trail to achieve regulatory traceability. This prospective framework offers a scaled solution to the problem of false positives to a significantly lesser degree, elevated true positive detection rates, and a defensible, efficient and adaptive sanctions screening program to meet contemporary regulatory standards by making use of the critical triad of accuracy, explainability and auditability.

Keywords: AI-Powered Sanctions Screening, Reduction of false positive, Explainable AI, Predictive Compliance, Model Risk Management, Financial Crime Technology, Regulatory Technology.

Paper Id: 232939

Published On: 2026-02-15

Published In: Volume 14, Issue 1, January-February 2026

Share this