Sanctions Screening
Sanctions screening is the process of checking customers, counterparties, and transactions against official sanctions lists, OFAC’s SDN List, the UN Consolidated List, EU and UK regimes, to identify prohibited dealings before or after they happen. A name check alone isn’t enough. OFAC’s 50 Percent Rule blocks entities that aren’t even named on any list, and a September 2025 Federal Reserve study found large language models cut screening false positives by 92% compared with standard fuzzy matching.
Key takeaways
- Sanctions screening checks against official government lists (OFAC, UN, EU, UK), unlike PEP screening, which relies on commercial databases.
- OFAC’s 50 Percent Rule blocks entities owned 50%+ by sanctioned persons automatically, even where the entity is never named on any list.
- The 50% OFAC blocking threshold is distinct from the 25% beneficial ownership threshold used in standard CDD; conflating the two is a common mistake.
- A September 2025 Federal Reserve working paper found LLMs cut sanctions screening false positives by 92% and improved detection by 11% versus fuzzy matching, though at far greater computational cost.
- OFAC applies strict liability: a firm can be penalised for a violation regardless of intent or awareness.
- OFAC doesn’t mandate a specific screening methodology, leaving firms to justify a risk-based approach.
- Real-time screening catches new prohibited transactions; batch re-screening catches customers who become sanctioned after onboarding, the more common scenario in practice.
On this page
What sanctions screening actually isWhy a name check alone isn’t enoughThe sanctions screening process step by stepReal-time screening vs batch screeningWhy fuzzy matching creates most false positivesWhat a 2025 Federal Reserve study foundStrict liability: why intent doesn’t matterScreening payments specifically: SWIFT, SEPA, and instant paymentsDocumenting alert decisions that hold upBuilding a sanctions screening programme that holds upFAQsRead more
92%
Reduction in sanctions screening false positives from LLMs versus standard fuzzy matching
50%
OFAC ownership threshold at which an unlisted entity is automatically blocked
Source: OFAC, 50 Percent Rule guidance
90%+
Commonly cited false positive rate in sanctions screening alert volumes
What sanctions screening actually is
Sanctions screening is the process of checking customers, counterparties, and transactions against official sanctions lists to identify prohibited dealings, either before a relationship is established or before a transaction settles. The core mechanism is a name and identifying-detail match against lists such as OFAC’s Specially Designated Nationals List, the UN Consolidated List, and EU and UK sanctions regimes.
Unlike PEP data, which comes from commercial vendors compiling public information, sanctions lists are official government or intergovernmental publications. That distinction matters: a firm can, in principle, screen directly against a primary source rather than relying entirely on a third party’s interpretation of who’s covered.
Why a name check alone isn’t enough
A screening programme that only checks whether a counterparty’s name literally appears on a list has a structural blind spot. OFAC’s 50 Percent Rule states that any entity owned 50% or more, in aggregate, by one or more SDN-listed persons is itself blocked, automatically, even though it never appears on the SDN List by name. OFAC doesn’t publish these unlisted-but-blocked entities; firms have to trace ownership structures themselves to find them.
This is genuinely different from the 25% beneficial ownership threshold used in CDD, and conflating the two numbers is a common mistake. The 25% figure identifies who has to be recorded as a beneficial owner for due diligence purposes. OFAC’s 50% figure is a separate, higher bar that determines whether an entity is already legally blocked, regardless of whether it was ever named.
The sanctions screening process step by step
A working sanctions screening process runs through a consistent sequence: match the customer or counterparty’s name and identifying details, date of birth, nationality, address, against relevant list entries; apply fuzzy matching to catch transliterations, misspellings, and known aliases rather than relying on exact string matches alone; trace beneficial ownership to check for 50 Percent Rule exposure where the direct counterparty isn’t listed itself; review and disposition any resulting alert, documenting why it was cleared or escalated; and re-screen on an ongoing basis, since lists update without a fixed schedule and a clean customer today can be designated tomorrow.
Real-time screening vs batch screening
Real-time screening checks a transaction or new relationship against current list data before it settles or completes, stopping a live payment or onboarding attempt involving a sanctioned party. Batch screening instead re-runs an existing customer base against updated lists periodically, catching a customer or counterparty who becomes sanctioned after the relationship was already established. Both matter for different reasons: real-time screening prevents new prohibited transactions; batch re-screening catches exposure that develops after the fact, which is the more common scenario in practice since most customers are screened clean at onboarding and only later designated.
Why fuzzy matching creates most false positives
Sanctions screening’s hardest operational problem isn’t missing genuine hits, it’s the volume of false ones. Fuzzy matching, deliberately set loose enough to catch a disguised, transliterated, or misspelled name, also flags large numbers of innocent customers who happen to share a name with, or closely resemble, someone on a list. Academic research on the subject has found false positive rates in sanctions screening commonly exceeding 90% of generated alerts, meaning the overwhelming majority of matches a system flags turn out, on review, to be nothing.
OFAC itself has acknowledged this dynamic directly: it recognises that certain weak aliases can generate a large volume of false hits, and its regulations don’t mandate one specific screening methodology, leaving financial institutions to make risk-based choices about their own matching approach rather than prescribing a single technical standard.
What a 2025 Federal Reserve study found
A Federal Reserve Board working paper published in September 2025, examining four families of large language models against common fuzzy matching algorithms for name and address similarity in a sanctions screening context, found that LLMs reduced false positives by 92% and increased detection rates by 11%, on average across realistic matching thresholds, compared with the best-performing traditional fuzzy matching baseline.
The same study is careful about a genuine trade-off: the LLMs tested were, on average, over four orders of magnitude slower than fuzzy matching methods. The authors proposed a model cascade instead, routing straightforward cases through fast fuzzy or exact matching and escalating only genuinely ambiguous cases to the slower, more accurate LLM review, noting that high-throughput environments like real-time payments likely need this tiered approach, while slower processes like account-opening due diligence could rely on LLM screening more directly.
Strict liability: why intent doesn’t matter
OFAC operates on a strict liability basis for sanctions violations. Unlike many compliance obligations that consider whether a firm knew or should have known about a problem, a sanctions violation can occur, and be penalised, regardless of intent. A firm that unknowingly transacts with a newly designated entity can still face civil penalties, which is precisely why continuous re-screening against current list data matters more here than in some other compliance areas: ignorance of a recent designation isn’t a defence.
Screening payments specifically: SWIFT, SEPA, and instant payments
Payment screening carries its own complexity because a single payment message contains multiple parties, sender, beneficiary, and sometimes intermediary banks or ultimate parties referenced only in free-text remittance information, each of which needs checking. Regulatory expectations from FATF and the Basel Committee call for screening in real time or near-real time, applied to both inbound and outbound payments, with list updates applied without delay.
OFAC issued specific guidance addressing faster and instant payment systems, making one principle explicit: faster settlement doesn’t reduce sanctions screening obligations. A payment that clears in seconds still needs to clear sanctions screening in that same window, which has pushed the industry toward exactly the kind of tiered, latency-aware screening architecture the Federal Reserve’s 2025 research describes.
Documenting alert decisions that hold up
Every sanctions screening alert, cleared or escalated, needs a documented, auditable rationale: what triggered the match, what information was reviewed to reach a decision, and who made the call. Given that OFAC’s strict liability standard doesn’t excuse an undetected violation, the quality of a firm’s screening decisions, and its ability to show its reasoning after the fact, matters as much as the technology generating the alerts in the first place.
Building a sanctions screening programme that holds up
A sanctions screening programme that holds up under regulatory review generally combines a few elements: screening against multiple official lists rather than a single consolidated source, since OFAC, UN, EU, and UK lists don’t fully overlap; beneficial ownership analysis built in to catch 50 Percent Rule exposure, not just direct name matches; a documented, risk-based matching methodology, since OFAC doesn’t prescribe one; and continuous re-screening rather than a check performed only at onboarding.
Trace ownership for 50 Percent Rule exposure
Check beneficial ownership alongside sanctions data to catch entities that aren’t directly listed.
Frequently asked questions
What is sanctions screening?
Sanctions screening is the process of checking customers, counterparties, and transactions against official sanctions lists, such as OFAC’s SDN List, the UN Consolidated List, and EU and UK regimes, to identify prohibited dealings.
What is OFAC’s 50 Percent Rule?
Any entity owned 50% or more, in aggregate, by one or more SDN-listed persons is itself automatically blocked, even if it never appears on the SDN List by name. OFAC doesn’t publish these entities; firms must trace ownership themselves.
Is the 50 Percent Rule the same as the 25% beneficial ownership threshold?
No. The 25% figure used in customer due diligence identifies who must be recorded as a beneficial owner. OFAC’s 50% figure is a separate, higher threshold that determines whether an unlisted entity is already legally blocked.
What is the difference between real-time and batch sanctions screening?
Real-time screening checks a transaction or new relationship before it completes, stopping a live prohibited payment. Batch screening re-runs an existing customer base against updated lists, catching customers who become sanctioned after onboarding.
Why does sanctions screening generate so many false positives?
Fuzzy matching, set loose enough to catch disguised or misspelled names, also flags large numbers of innocent customers who share a name with or resemble someone on a list. Academic research has found false positive rates commonly exceeding 90% of alerts.
What did the 2025 Federal Reserve study on sanctions screening find?
Testing four families of large language models against standard fuzzy matching, the study found LLMs reduced false positives by 92% and increased detection rates by 11%, though the models were over four orders of magnitude slower, leading the authors to propose a tiered cascade approach.
Does OFAC require a specific sanctions screening method?
No. OFAC’s regulations don’t mandate a specific screening methodology, leaving firms to make risk-based decisions about their own matching approach and technology.
Does intent matter for a sanctions violation?
No. OFAC operates on a strict liability basis, meaning a firm can be penalised for a violation even if it didn’t know, and had no way of knowing, that a counterparty had been designated.
Read more: our ultimate guides, whitepapers and templates
Related guides and resources to help you act on what you just read.
Last reviewed July 19, 2026 · 11 min read · Written for compliance and risk professionals · By the WhoWiki editorial team
Key takeaway: Sanctions screening is the process of checking customers, counterparties, and transactions against official sanctions lists, OFAC’s SDN List, the UN Consolidated List, EU and UK regimes, to identify prohibited dealings before or after they happen. A name check alone isn’t enough. OFAC’s 50 Percent Rule blocks entities that aren’t even named on any list, and a September 2025 Federal Reserve study found large language models cut screening false positives by 92% compared with standard fuzzy matching.