Sanctions Screening - OFAC, EU, UK, UN Bulk Lists

Screen a counterparty name against the four government sanctions lists that compliance teams actually get asked about: OFAC SDN, the EU consolidated list, UK OFSI and the UN Security Council consolidated list. One run returns matched rows with the list source, programme, reference id, date of birth and country where the list records them, plus a single merged verdict per name. Built for marketplaces vetting new sellers and fintechs screening counterparties before payout, invoice finance or onboarding.

What you get

• One dataset row per matched list entry, carrying list_source (ofac_sdn, eu_fsf, uk_ofsi, un_sc), programme, reference_id, entity_type, entity_name, aliases, dob, country and addresses.
• A merged verdict per screened name: hit, possible_hit or no_hit, with lists_hit naming which of the four lists carry the entry.
• Match diagnostics on every row: match_type (exact, normalised, fuzzy), score from 0 to 1, and the flags dob_conflict, country_match and lei_match so you can see why a row survived or got dropped.
• Batch screening: put one name per line in names and get rows for the whole list in one run.
• Local matching on a SQLite index built from the bulk files at run start, so the screen itself does not depend on a third-party search API being up.
• A summary record in the key-value store with per-name verdicts, row counts per list, and when the bulk files were last fetched.

Why sanctions screening

Compliance budgets buy certainty, not dashboards. The finding that changes a decision is binary: counterparty Y is named on OFAC SDN under programme Z, so the account is blocked and the case is escalated. A risk index, a coverage score or a comparative metric does not trigger that action. What teams need is the actual list row, with the programme and the reference id their auditor will ask for, and a clear statement of which lists were searched.

The four lists we screen are the ones a regulator names when they ask "did you check?". OFAC SDN is the US list with secondary-sanctions reach. The EU consolidated list is what a Dutch or German payments firm needs. UK OFSI covers the UK regime after Brexit. The UN Security Council list is the floor almost every national regime builds on. Screening all four in one call is the difference between a real check and a checkbox.

Input

Field	Required	Description
name	yes	Person or legal entity name as the counterparty gave it. Prefilled with Deripaska so the form runs as-is.
names	no	One name per line, for batch screening of a seller list.
country	no	ISO code or country name. Demotes fuzzy matches from other countries; exact name hits survive.
dob	no	YYYY-MM-DD or any text carrying a year. A conflicting year removes the match.
lei	no	Legal Entity Identifier, checked against list reference identifiers.
lists	no	all, ofac_sdn, eu_fsf, uk_ofsi or un_sc. Defaults to all four.
match_threshold	no	Similarity floor for fuzzy matches, 0.7 to 1.0. Default 0.88. Exact and normalised matches always pass.
refresh_lists	no	Force a re-download of the bulk files even if the cache is under 24 hours old.

Output

{
  "query_name": "Deripaska",
  "verdict": "hit",
  "lists_hit": ["eu_fsf", "ofac_sdn", "uk_ofsi"],
  "match_name": "Oleg VLADIMIROVICH DERIPASKA",
  "match_type": "normalised",
  "score": 0.96,
  "list_source": "ofac_sdn",
  "programme": "RUSSIA-EO14024",
  "reference_id": "16152",
  "entity_type": "individual",
  "entity_name": "Oleg Vladimirovich DERIPASKA",
  "aliases": ["Oleg Vladimirovich DERIPASKA"],
  "dob": "1968-01-02",
  "country": ["RU"],
  "addresses": ["Moscow, RU"],
  "dob_conflict": false,
  "country_match": true,
  "lei_match": null
}

Names with no match return a single row with verdict: "no_hit" and empty match fields, so a clean screen is visible in the dataset and not just an absence of rows.

How matching works

Names are normalised (diacritics folded, punctuation and legal suffixes such as Ltd, LLC, GmbH and PJSC stripped) and matched in three tiers. exact means the raw strings agree. normalised means they agree after folding. fuzzy means a token-set overlap or character similarity above the threshold. A date of birth on the query that contradicts the listed year removes the row outright. A country on the query that contradicts every listed country drops a fuzzy candidate but keeps an exact name hit, because lists record nationality loosely and a missed hit costs more than a false one.

Use cases

Marketplace seller onboarding. Before a new seller goes live, screen the legal name and the beneficial owner against all four lists. A hit on any one of them is a block and an escalation, and the dataset row gives compliance the programme and reference id to file. Batch mode takes the weekly seller queue in one run.

Fintech counterparty checks before payout. Invoice finance and payout flows need a check at the moment money moves. Run the payer name, the country and the date of birth as provided; the DOB filter keeps common names from producing noise on the exact-match path, and verdict is the field your decision rule keys on.

Periodic rescreening. Lists change. Rescreen your book of counterparties on a schedule and diff reference_id and programme against the last run to see who was newly designated or delisted.

Audit evidence. The dataset is a dated record of who you screened, against which lists, with what result. That is the artefact an auditor asks for when they ask how you satisfied the sanctions screening requirement.

How it compares

OpenSanctions aggregates these lists and hundreds more, and sells access under a commercial data licence with per-customer terms. Sanctions.io and the KYB screening actors on this platform sell a similar check as a subscription or per-lookup product. What differs is the shape: this actor returns the raw government list rows with their own reference ids and programmes, priced per screening event, with no seat licence and no monthly minimum. If you need politically exposed persons, adverse media or a curated deduplicated entity graph, a commercial dataset is the right tool and this actor is not a substitute for it.

Pricing

$0.025 per screening event: one event per name screened, whether the result is a hit or a clean no_hit. Batch runs charge one event per name in the batch. All pricing is pay-per-event, you only pay for results you receive. No actor-start fee, no per-compute-unit charges.

Limits and gotchas

• Politically exposed persons are out of scope. PEP coverage requires a commercial data licence we do not bundle; the underlying aggregators publish under non-commercial terms with serious contractual penalties for resale. This actor screens sanctions lists only.
• The first run in a fresh environment downloads the bulk files (tens of megabytes) and builds the SQLite index, which adds time before the first result row. Later runs in the same environment reuse a cache younger than 24 hours; refresh_lists forces a rebuild.
• Fuzzy matching is name-based. Misspelled transliterations of Cyrillic or Arabic names can score below the threshold: lower match_threshold and review possible_hit rows by hand.
• Date of birth on many list rows is partial (year only, or absent). A DOB filter only removes a row when both sides carry a year and the years disagree.
• Vessels and companies appear alongside individuals; check entity_type before applying a person-only workflow.
• The four lists are the scope. National regimes beyond the EU and UK (Swiss SECO, Canadian consolidated, Australian DFAT) are not searched.
• A no_hit is a statement about these four lists on the day of the run. It is not a legal opinion and not a substitute for your own compliance policy.

FAQ

Do I need an API key for the government lists? No. The four bulk files are published for direct download. The actor fetches them itself and matches locally.

Can I screen a whole seller list in one run? Yes. Put one name per line in names. Each name is screened separately and charged as one screening event.

What does possible_hit mean? A name that matched above the fuzzy threshold but not exactly. Review those rows before acting; they are the ones a human should look at.

Does a date of birth help with common names? Yes. Supplying dob removes candidates whose listed birth year disagrees, which is the main source of false positives on names like Smith or Ivanov.

Why is PEP screening not included? Politically exposed persons data comes from commercial aggregators whose licences forbid resale. Bundling it would need a reseller arrangement we do not hold, so it is explicitly out of scope.

Related Actors

• https://apify.com/s-r/market-quotes
• https://apify.com/s-r/cost-of-living