US, UK and EU Regulators Set AML Transaction Monitoring Expectations

US banks must file suspicious activity reports within 30 days, the FCA has issued monitoring guidance, and the EU’s AML rulebook applies from July 10, 2027.

US, UK and EU Regulators Set AML Transaction Monitoring Expectations

Banks buying AML transaction monitoring software face three rulebooks: US suspicious activity reports due within 30 days, UK Financial Conduct Authority guidance issued in November 2024, and EU rules applying from July 10, 2027.

The rules do not describe the same thing in the same way, but they ask a similar question of every bank: can it show that the system flagging customer activity is calibrated to its own risks, tested, explainable and tied to a working investigation process. For compliance teams and the executives who fund them, that question decides whether a platform is a control or only a cost.

How AML transaction monitoring software works end to end

The process starts with data. Monitoring software ingests transactions from core banking, payments and card systems, together with customer and account data from know-your-customer (KYC) files, and prepares it for analysis. Oracle’s documentation for its Transaction Monitoring product describes this stage as filtering, validating, deriving, aggregating and joining data, with data quality checks applied before any detection runs.

Detection comes next. Under rules-based monitoring, scenarios such as rapid movement of funds or cash deposits just under a reporting threshold carry parameters, and a transaction that crosses them generates an alert. Machine learning models work differently, scoring activity against a customer’s own history and that of peer groups, so they can surface behavior no written rule anticipates. NICE Actimize’s Suspicious Activity Monitoring (SAM) product, for example, uses an anomaly model that compares entities and transactions with peer and historical activity, and ranks alerts by predicted risk.

Alerts then move to alert triage, where an analyst closes clear false matches and escalates the rest into case management. An investigator gathers account history, counterparties and documents, and a compliance officer decides whether to file a suspicious activity report (SAR) with the national financial intelligence unit. Timing of detection also matters, and real-time versus batch AML monitoring is covered in a separate RegTech Post analysis. The core idea of transaction monitoring is the same either way.

Where false positives in AML transaction monitoring software come from

Most false positives trace back to three sources. Thresholds set too low, or copied from vendor defaults without reference to the bank’s customer base, flag ordinary behavior such as a business paying its suppliers in round amounts. Poor data is the second cause: missing customer occupation, stale risk ratings or unlinked accounts leave a model without the context needed to see that an alert is normal. The third is scenario sprawl, where rules accumulate over years and overlap, so one event produces several alerts.

The FCA treats this directly. Its Financial Crime Guide lists as poor practice a threshold-based, rule-driven system that is poorly calibrated and whose rationale the firm cannot explain. That makes tuning a regulatory matter as well as a cost matter. RegTech Post’s transaction monitoring tuning template pack sets out a structure for documenting threshold changes and their results.

What US regulators expect of AML transaction monitoring

US banks must file a SAR under 31 CFR 1020.320, part of the Bank Secrecy Act rules enforced by the Financial Crimes Enforcement Network (FinCEN). The filing is due no later than 30 calendar days after the bank first detects facts that may form a basis for a report. If no suspect is identified, the bank may take another 30 days to find one, but never more than 60 days in total. Reporting thresholds are generally $5,000 where a suspect can be identified and $25,000 regardless of a suspect.

FinCEN, together with the Federal Reserve, FDIC, National Credit Union Administration and Office of the Comptroller of the Currency (OCC), published SAR frequently asked questions on October 9, 2025. They state that banks are not required to document a decision not to file a SAR, although FinCEN encourages a short, concise record. They also say a bank need not run a separate review after filing to see whether suspicious activity continued, and that filing continuing-activity SARs every 90 days is an option rather than a requirement. The FAQs add that monitoring should match an institution’s money laundering and terrorist financing risk, and that appropriately tailored automated systems may be used.

Model risk is the second US pillar. On April 17, 2026, the OCC, the Federal Reserve and the FDIC replaced their 2011 guidance with OCC Bulletin 2026-13. The bulletin also rescinds OCC Bulletin 2021-19, the earlier statement on model risk management for BSA and AML systems. It defines a model as a complex quantitative method that applies statistical, economic or financial theories, and excludes deterministic rule-based processes that use no such theory, so a pure rules engine may fall outside it while a machine learning scorer sits inside it. Generative and agentic AI models are outside the scope, and the guidance says it “does not set forth enforceable standards or prescriptive requirements.”

For buyers, the practical result is that model validation of any statistical scoring remains an exam topic, covering conceptual soundness, outcomes analysis and vendor products. RegTech Post’s model risk management template pack covers the documentation that validators ask for.

UK and EU rules on AML transaction monitoring software

In the UK, the FCA added transaction monitoring material to its Financial Crime Guide in November 2024 through Policy Statement PS24/17. It applies to manual and automated monitoring alike, and it includes self-assessment questions, examples of good and poor practice, and a case study drawn from weaknesses the regulator found in HSBC’s monitoring. Among the good practices, new approaches are piloted or run through evaluation periods so the firm can show appropriate testing.

The EU is building a single rulebook. Regulation (EU) 2024/1624, the AML Regulation (AMLR), applies directly in all member states from July 10, 2027. It requires obliged entities to scrutinize transactions throughout a customer relationship and to report suspicions to the national financial intelligence unit. The Anti-Money Laundering Authority (AMLA), set up under Regulation (EU) 2024/1620, is developing technical standards and guidance, and RegTech Post has set out what AMLA means for compliance data. Banks operating across the three jurisdictions will in practice need one monitoring platform that can hold separate scenario sets and reporting formats for each.

Checklist for buying AML transaction monitoring software

Five questions cover most of what examiners and boards ask about AML transaction monitoring software.

  • Tuning: Can compliance staff change thresholds and test the effect on historical data before going live, and is every change versioned? Napier AI’s platform offers a sandbox that runs on real data and keeps version history automatically, while Flagright offers shadow mode and 90-day backtesting. Oracle’s scenarios carry threshold parameters that users can adjust, and customers can build their own scenarios.
  • Explainability: Can an analyst see why an alert fired, in plain language, and can the bank explain a model’s output to an examiner? Napier AI generates separate scores for its rules and its AI, and describes what drives each alert. Hawk’s transaction monitoring product is built around explainable AI aimed at reducing unnecessary alerts.
  • Model validation: Does the vendor supply development documentation, performance testing and a path for independent validation? Under OCC Bulletin 2026-13, vendor products fall within validation expectations when they use statistical models.
  • Case management: Do alerts, investigations, decisions and SAR drafts sit in one workflow with a full audit trail? Napier AI’s guided case management and Flagright’s automated case management product both aim at this, and NICE Actimize SAM covers detection through workflow and reporting.
  • Core banking integration: Does the software take data from core banking through supported connectors or APIs, and can it run on premises or as software as a service? NICE Actimize SAM is offered both ways, and Napier AI deploys hosted or on premises. Flagright publishes a 200 millisecond p99 API latency for real-time use. Oracle’s Financial Crime Compliance Management suite has been integrated with FLEXCUBE core banking at Abay Bank.

Cost belongs on the list as well, because investigation time drives most of the spend. Napier AI has published an estimate of $143 billion in untapped AI savings for AML, and newer entrants such as Flagright, which raised a $12.5 million Series A, compete on faster deployment and lower tuning effort against established suites from Oracle and NICE Actimize.

What to watch next for monitoring teams

US banks are working to the revised model risk guidance, which the agencies say they will follow with a request for information on model risk management, including banks’ use of AI. In the UK, the FCA’s guidance sits in the Financial Crime Guide for firms to measure against at their next review. In the EU, the AMLR applies from July 10, 2027, the date by which banks need monitoring scenarios, data and reporting aligned to a single rulebook.