FAIS — FinCEN's Pioneering AI System for Detecting Money-Laundering Leads from Cash-Transaction Reports

The U.S. Bank Secrecy Act requires banks to file a Currency Transaction Report (CTR) for cash transactions over $10,000, generating a volume of paperwork no team of human analysts could review manually. The Financial Crimes Enforcement Network (FinCEN) built FAIS, an artificial-intelligence system combining rule-based reasoning with a shared "blackboard" database, to link CTRs, businesses and individuals and flag patterns consistent with structuring or layering.

FAIS went live in March 1993. A dedicated group of analysts used it to process roughly 200,000 transactions a week.

By the time of a 1995 peer-reviewed case study in AI Magazine, the system had generated more than 400 investigative support reports corresponding to over $1 billion in potential laundered funds — leads that were handed to law-enforcement agencies for further investigation. A contemporaneous review by the U.S. Congress Office of Technology Assessment corroborated the system's operational status and role in identifying leads that would not otherwise have surfaced from the raw filings.

The documented evidence dates to the early-to-mid 1990s and describes an internal law-enforcement tool: the sources reviewed do not include a more recent independent evaluation, nor do they describe a public transparency or audit mechanism. FAIS predates modern machine-learning methods, relying instead on expert rules — an early but rigorously documented example of AI applied to a public-sector compliance function.