Are AI-Based Anti-Money Laundering Systems Compatible with Fundamental Rights?

Abstract

Anti-money laundering and countering the financing of terrorism (AML/CFT) systems must comply with the GDPR and the proportionality test under European fundamental rights law, as most recently expressed by the Court of Justice of the European Union (CJEU) in the Digital Rights Ireland and Tele2 Sverige - Watson cases. The objective of this paper is to present how AML/CFT laws and systems work, how artificial intelligence (AI) can enhance those systems, and examine whether these systems comply with the European proportionality test. We conclude that current AML/CFT systems violate the proportionality test in several ways: AML/CFT laws are not specific enough to satisfy the ‘provided by law’ test, and the lack of feedback on the utility of suspicious activity reports sent by banks to law enforcement authorities means that it is impossible to determine if AML/CFT systems are ‘genuinely effective’. We propose a scoring mechanism that would permit banks to receive feedback from law enforcement authorities on what alerts are ‘genuinely effective’ and adjust their monitoring systems to be more proportionate. We also propose (i) that banks be required to inform their customers when they have been targeted by a suspicious activity report, as soon as doing so would no longer compromise an investigation, and (ii) the creation of dedicated institutional oversight on fundamental rights aspects of AML/CFT, similar to what exists for intelligence gathering.

Keywords: AML, anti-money laundering, AI, artificial intelligence, fundamental rights, proportionality