Real-time fraud detection and risk management - autonomous AI agents
Abstract
Synthetic identities created by fake real or fabricated data and AI generated Attacks orchestrated by AI are the new and emerging threats, which are not detectable by Standardized systems as they are not able to quickly adapt them. In real time AI Agents by explaining and performing adaptive risk management promising to improve the detection of Financial Fraud very accurately and instantaneously. We propose a Modular Agentic AI system Explainable AI for compliance, Federated Learning for privacy and Reinforcement Learning for adaptability. Experiments on synthetic and benchmark datasets demonstrate sub-100ms response times and
Fields of Study & Subjects
Imbalanced Data Classification Techniques
Privacy-Preserving Technologies in Data
Spam and Phishing Detection
Publication details
- Journal / source
- IET conference proceedings.
- Volume
- 2025
- Issue
- 43
- Pages
- 199–205
- Publication date
- 1 February 2026
- Language
- EN