Every list change, dated and sourced, the morning it happens

In practice the list changes on most working days, not on a fixed schedule. Across the twelve months to 18 September 2026 the Monitor recorded changes on 214 of 250 US business days, a median of six party level changes per publishing day. Screening on a monthly cycle therefore leaves an average exposure window of roughly fifteen business days between a designation and your first screening run against it.

It requires a documented AML and CFT control framework mapped to named standards, with evidence that each control operates. The load falls on four areas: customer risk rating at onboarding and on review, sanctions screening against current list versions, transaction monitoring with tuned and tested thresholds, and reporting to the NFIU within the statutory window. Institutions are expected to evidence dates, list versions and approvals, not simply to hold a policy.

A currency transaction report is threshold driven and mechanical: it is filed because a cash transaction crossed a reporting amount, regardless of suspicion. A suspicious transaction report is judgement driven: it is filed because the institution has grounds to suspect the funds or the conduct, whatever the amount. A CTR needs the transaction record. An STR needs the reasoning that led to suspicion, recorded at the time.

The obligations are the same in kind but different in shape. A payment service provider usually carries higher volume, lower value and a merchant book rather than a retail deposit book, so weight moves from customer due diligence at account opening to merchant onboarding, ongoing merchant risk and velocity monitoring. Beneficial ownership on merchants becomes the harder problem, and settlement exposure replaces cash handling as the dominant risk.

A general licence permits a class of transaction that would otherwise be prohibited, without an application. Papering against one means recording the licence number, the exact clause relied on, the dates the activity falls within, and the evidence that the transaction meets every condition. Most carry a reporting obligation, commonly within fourteen or thirty days. Keep the licence text as it stood on the transaction date, because licences are amended and superseded.

Treat it as an expiry, not an exoneration. Where the national emergency or determination carrying a designation lapses, the record comes off without any finding that the original grounds were wrong. Unblock where you are required to, but retain the evidence file, the decision note and the transaction history, and keep the party on internal review. Redesignation under a replacement instrument does not require new conduct.

Ask for evidence rather than claims: list coverage with update frequency and publishing lag per source, screening performance at your book size, false positive rates measured on analyst dispositions rather than estimated, NFIU filing formats supported end to end, audit trail and model change control, data residency, and a named implementation team with reference deployments at comparable volume. Ask every vendor to answer against your own book.

Run a backtest on your own settled traffic, not a sample. Measure alerts per thousand transactions, precision confirmed by analyst disposition, time from event to alert at peak, and behaviour during a replay of your busiest day. Ask for the tuning method and who signs off a threshold change. A vendor that cannot show precision measured on dispositions, separately from precision estimated on unreviewed alerts, has not measured it.