Lynx Tech launches Adaptive AI to tackle rising fraud

The technology is being introduced first in the United States, with European institutions to follow. The models are designed to integrate within two weeks and become fully optimised within three to six months, learning continuously from new data without requiring manual updates. 

According to UK Finance, fraud incidents are growing by 16%, with losses exceeding GBP 3 million per day. In the context of this trend, banks are increasingly relying on AI-driven tools to detect and prevent sophisticated financial crimes in real-time. 

The updated DAMs from Lynx are positioned to address challenges that static Machine Learning (ML) models face, such as degradation due to data drift. By retraining daily and supporting data extensibility, the new models aim to offer sustained detection rates and reduce false positives, particularly in identifying money mule activity and complex fraud networks.

Shift from transaction analysis to network-based detection

The models are structured to move beyond individual transaction monitoring and instead focus on uncovering larger fraud networks. This involves analysing interconnected behaviours across multiple accounts to expose previously hidden fraudulent schemes. 

Representatives from Lynx indicated that the new system could analyse up to 40,000 data points in real time. The models are based on in-memory databases and customised algorithms, developed over two years of research. This infrastructure reportedly supports a fast integration and real-time detection without adding operational downtime. 

Officials from Lynx stated that the improvements provide a more nuanced understanding of financial behaviours across entire account networks. This has contributed to a noticeable drop in false positives and more accurate identification of fraud attempts. 

A whitepaper detailing the technology’s development and deployment has also been published by the company. A Lynx official noted that as financial institutions prepare for the UK’s revised APPF reimbursement rules, adaptable AI systems such as these may become critical tools in minimising exposure to fraud-related losses.