Japan’s FSA highlights Basel Committee guidance on AML data collection and AI transaction monitoring

Japan’s Financial Services Agency has published a notice drawing attention to a new Basel Committee newsletter on anti-money laundering and counterterrorism financing supervision, highlighting two concrete areas of focus for banks: more structured collection of AML risk data across jurisdictions and closer scrutiny of artificial intelligence tools used in transaction monitoring.

The FSA notice, published Oct. 28, said the Basel Committee on Banking Supervision released its “AML/CFT supervisory newsletter” on Oct. 19. The document is BCBS Newsletter No. 37. In its notice, the FSA said: “バーゼル銀行監督委員会（以下、『バーゼル委』）は、10月19日、『マネー・ローンダリング及びテロ資金供与対策（AML/CFT）に関する監督ニューズレター』（原題：AML/CFT supervisory newsletter）を公表しました。”

The development matters because the Basel Committee is the main global standard-setting body for prudential bank regulation and supervision. Its publications are not binding law on their own, but they often influence how national regulators shape supervisory expectations. In that sense, the newsletter offers a window into the issues bank supervisors are actively examining in AML/CFT programs.

According to the BIS and Basel Committee publication summary, the newsletter has two main components. One is a supervisory stocktake across 19 jurisdictions on AML/CFT data collection and risk-assessment methods. The other is a summary of an industry workshop on the role of AI and machine learning in transaction monitoring systems, which banks use to flag potentially suspicious activity.

The BIS summary said the publication “summarises key themes and challenges discussed in recent supervisory exchanges and workshops related to Anti-Money Laundering and Countering the Financing of Terrorism (AML/CFT).” On the data side, the stocktake found that many authorities rely on AML/CFT questionnaires and combine that information with multiple other sources, including financial intelligence units, audits and transaction data.

The exercise also pointed to the increasing scale and granularity of supervisory data collection. Some authorities typically collect more than 100 data points, according to the BIS summary. That suggests supervisors are seeking a more structured and comparable view of banks’ AML/CFT risk profiles, particularly where operations span multiple markets.

At the same time, the stocktake identified recurring problems. Among them were manual processing, weak data quality, cross-border complexity and the challenge of weighing a bank’s inherent risk — the exposure it faces based on its business and customers — against the effectiveness of its controls.

The second half of the newsletter turns to AI and machine learning in transaction monitoring. The BIS summary said supervisors are interested in the technology’s potential use in detecting unusual or suspicious patterns, but it also underscored governance and explainability concerns. Those issues go to whether banks can properly oversee such systems and clearly explain how alerts or decisions are generated.

The newsletter itself does not create new legal requirements. Still, the FSA’s decision to republish the Basel item underscores its relevance for banks under supervisory scrutiny in Japan, especially on AML/CFT data quality, reporting practices and the governance of AI-based monitoring tools.