AI glossary · Basic · Classic ML, recsys & search · Data engineering · Finance & insurance
AML
AML (Anti-Money Laundering) refers to the rules and technologies through which financial institutions verify customers, find suspicious transactions and report them to authorities, in order to prevent money earned from crime from being laundered to look like legitimate money.
About this term
- Level
- Basic · Technical terms you often meet in AI news
- Category
- Industry AI
- Industries
- Finance & insurance
- Added
- 2026-10-10
How it works
First, when an account is opened, the institution verifies who the customer is (KYC) and assigns a risk level. After that, it keeps monitoring transactions and raises an alert when it sees suspicious patterns, such as splitting money into small deposits over a short period or moving funds quickly through several accounts. A staff member investigates the alert, and if they judge it to be genuinely suspicious, they report it to the financial authorities as a suspicious transaction. The institution also screens customers and counterparties against sanctions lists.
From transactions to suspicious transaction reports
What AI does
Traditional AML systems often relied on rules set by people, so they produced a very large number of false positives, where alerts went off for legitimate transactions. Machine learning can score the risk of alerts to set the order of investigation, or use anomaly detection to find new patterns that are hard to catch with rules. Viewing and analyzing the flow of money between accounts as a graph helps uncover organized laundering that links many accounts together. Recently, LLMs have also been used to summarize investigation materials and draft reports.
Things to keep in mind
AML is a legal obligation, so financial institutions that neglect it can face heavy penalties. For this reason, decisions made by AI must also be explainable to auditors and regulators. Actual money laundering cases are very rare among all transactions, so class imbalance is severe, and correct labels are hard to obtain. Criminals also change their methods to evade detection, so models need to be updated continuously.
Example
A bank's transaction monitoring system looks for suspicious patterns in the transactions that pile up each day and generates alerts, and compliance staff investigate the alerts with the highest AI-assigned risk scores first. This reduces time spent on false positives and lets them focus on the truly risky cases.