ARTIFICIAL INTELLIGENCE-ENABLED REGTECH FOR KYC, FINANCIAL COMPLIANCE, AND REGULATORY REPORTING: APPLICATIONS, RISKS, AND FUTURE DIRECTIONS

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Financial institutions face growing volumes of customer, payment, ownership, and regulatory data while remaining accountable for risk-based compliance decisions. Artificial intelligence-enabled regulatory technology can assist this work by combining digital identity services, entity resolution, rules, machine learning, graph analytics, natural language processing, and workflow automation. This narrative review explains the technical architecture of such systems and examines their use in know-your-customer procedures, financial compliance monitoring, investigations, and regulatory reporting. The technology can improve data reconciliation, prioritise alerts, reveal relationships that are difficult to detect in isolated records, retrieve evidence, and automate repeatable reporting steps. Its value, however, is constrained by incomplete labels, severe class imbalance, concept drift, adversarial adaptation, opaque models, privacy risks, fragmented rules, legacy systems, and dependence on external vendors. A lower false-positive rate does not by itself demonstrate improved monitoring effectiveness, and a suspicious transaction or activity report should be understood as an indicator for further review rather than definitive evidence of non-compliance. Responsible implementation therefore requires a layered architecture in which traceable controls preserve explicit regulatory requirements, models are validated against operational objectives, consequential decisions receive documented human oversight, and every data transformation, model version, explanation, override, and report remains auditable. Future development is likely to centre on privacy-preserving collaboration, temporal graph models, machine-readable regulation, and source-grounded generative assistants. AI-enabled RegTech should be treated as governed decision support, not as an autonomous substitute for legal interpretation, customer due diligence, investigation, or regulatory accountability.