Japanese banks to share AML/CFT information

Among the purposes is to tackle false positives which it current systems range from 70% to 95%, Hiroshi Ozaki, director of the AML/CFT policy office of the country’s Financial Services Agency (FSA), said at an industry conference. Trials of a possible system have proved to be successful.

Plans to introduce one date back to revised guidelines the regulator issued in 2018 which highlighted the latest technology’s ability to increase effectiveness and efficiency of AML/CFT systems.

“New technologies such as AI (artificial intelligence), blockchain, and RPA [robotic process automation] are used in various phases of AML/CFT to improve the effectiveness of controls,” sector journal Regulation Asia quoted the guidelines as saying.

A proof-of-concept project last year used a shared platform to ingest transaction data from financial institutions, without sharing or pooling the data, and then analysed it using machine learning and assigned true positive scores to each transaction, Ozaki said.

The AI model deployed was able to learn from the datasets and be tuned to improve accuracy. The project was a success at triaging alerts, reducing false positives and enhancing efficiency in monitoring and screening, he added.

The Japanese Bankers’ Association is now working out how to implement the system across the banking industry.

In his address, Ozaki accepted costs are increasing for investing in and maintaining the systems, as well as integrating them with legacy systems, and supervisors have to pay attention to that pain point.