Congress is facing pressure to pass an Anti-Money Laundering Reform bill in the face of the recent FinCEN Files, an investigation into thousands of leaked suspicious activity reports (SARs) that financial institutions submitted to the U.S. government.
The investigation’s revelation that criminals are able to exploit the U.S. banking system to conduct criminal activities, money laundering, human and drug trafficking—is relatively unsurprising.
It’s akin to the scene in "The Music Man" where the townspeople of River City worry there will be gambling in the pool hall. What is newsworthy about the investigation is its conclusion that these very safeguards, which are designed to prevent crime, aren’t effectively addressing it. The system works as designed, but its design is so flawed that it actually facilitates criminal enterprises.
It's important that policymakers considering reforms to the AML regime recognize that current banking regulations attempt to balance two things that we perceive to be dichotomous: national security and privacy. In order to fight financial crime, bankers must identify money launderers and traffickers, and yet although it's not precluded, the legal framework actually limits the sharing of data about potential suspicious activity and customers between institutions.
U.S. law requires banks to develop AML programs that detect and report on suspicious activity, which generally means that banks hire former law enforcement officials to assist them in fulfilling these requirements. However, these expectations also encourage criminals to leverage multiple financial institutions to conduct their nefarious activity. For example, if one bank closes an account, criminals benefit from the lack of information sharing by simply repeating their activity at a different bank.
Not only does the presumption that national security and privacy are mutually exclusive create a system that criminals exploit, it’s also outdated—because its authors never imagined machine learning. Today, artificial intelligence and machine learning technologies allow institutions like banks, government agencies, and more, to collaborate for security purposes without actually sharing individuals’ private data.
Mathematical models informed by data can be shared collectively between organizations, in the cloud for example, without sharing the data itself. Collaboration with privacy protections will revolutionize crime-fighting.
