Private beta

Know who you’re about to wire money to — before you wire it.

SanctionScreen is a per-call sanctions screening API for humans and AI agents. Screen any name against OFAC and UK sanctions lists in ~110 ms, enrich any company domain, and get a cited answer your compliance review can actually defend.

Get a free API key — 50 calls/mo, no card See how it works
25,719 parties indexed
OFAC + UK FCDO lists, refreshed daily
Effectively $0
screening cost per call — no bill surprises
Cited, every time
list publish date + fixed disclaimer on each answer

How it works

Three steps. No seats, no enterprise contract, no sales call.

1

Call one tool.

Your agent (or your code) calls screen_counterparty with a name — optionally a country, date of birth, or ID numbers — or enrich_company with a domain.

2

We match, score, and cite.

25,719 sanctioned parties → 150,704 name variants, both name orderings, non-Latin scripts transliterated — matched with calibrated confidence tiers. Every hit returns the list, the record ID, the publish date, and a source citation.

3

You get a disposition, not a verdict.

clear, potential_match, or review_advised — with human review required before any adverse action. A potential match is the start of your review, never its conclusion.

The two tools

Plain language. Full schemas and few-shot examples live in the agent docs.

screen_counterparty

Sanctions screening

“Is this name on a sanctions list?” Give it a name (and anything else you know — country, date of birth, ID numbers). Back comes a disposition, every match above a low threshold with confidence scores, and a full citation per match.

  • All matches always returned, even low-confidence ones — hiding a weak match is a compliance liability.
  • Adding an identifier (like a date of birth) turns an exact-name match from medium into high confidence, raising the disposition to review_advised.
  • Language rule we follow: “potential match,” never “sanctioned” or “is on a list.”
enrich_company

Company enrichment

“Who is this company, really?” Give it a domain or a company name. Back comes legal name, industry, size band, employee range, HQ country and city, founding year where available — and a per-field source list.

  • Every field traced to its source. No black-box attributes.
  • v1 built on free public company-identity data — enrichment stays cheap.
  • Structured JSON your agent can parse directly.

Sample output

Structured, cited, and honest about uncertainty.

{
  "screening_id": "cs_scr_01J9…",
  "disposition": "potential_match",
  "matches": [
    {
      "list": "OFAC_SDN",
      "list_date": "2026-09-23",
      "record_uid": "35096",
      "matched_name": "PUTIN Vladimir",
      "name_kind": "alias_strong",
      "confidence": "medium",
      "confidence_score": 70,
      "matched_fields": ["name"],
      "citation": {
        "source": "US Treasury OFAC SDN List (SDN_ADVANCED.XML)",
        "url": "https://sanctionslistservice.ofac.treas.gov/api/PublicationPreview/exports/SDN_ADVANCED.ZIP",
        "accessed_at": "2026-09-26T00:00:00Z"
      }
    }
  ],
  "disclaimer": "Informational only — not a legal verdict. Not a substitute for due diligence. Human review required before any adverse action.",
  "generated_at": "2026-09-26T01:00:00Z"
}

Illustrative example — the shape of a real response, values not from a live call.

Pricing

Per call. No seats. MCP access included in every tier at no extra charge.

Free Proposed

$0/mo
  • 50 calls/mo total
  • Screens + enrichments
  • No card required
Pay as you go

PAYG Proposed

$0.10/screen
  • Screen $0.10 / call
  • Enrich $0.15 / call
  • Bundle $0.20 / call
  • $5 minimum top-ups

Starter Proposed

$29/mo
  • 300 screens included
  • 100 enrichments included
  • Overage at PAYG rates

Growth Proposed

$99/mo
  • 3,000 screens included
  • 1,000 enrichments included
  • Lower overage rates

Enterprise Proposed

Custom
  • SSO / SCIM
  • Audit logs
  • Zero-data retention
  • Static egress IP, SLA
Proposed pricing — not final. We only bill successful calls. Failures are never billed. Final numbers are confirmed before launch.

A potential match is never a legal verdict.

Every SanctionScreen response carries this fixed disclaimer, verbatim:

“Informational only — not a legal verdict. Not a substitute for due diligence. Human review required before any adverse action.”
  • We say potential match, never “sanctioned” or “is on a list.” Sanctions lists contain names, not verdicts — deciding what a name means is your call, with human review.
  • Every response carries list_date — the publish date of the list file it was matched against — so you always know how fresh the underlying data is.
  • All matches above a low threshold are returned, including low-confidence ones. Screening silently is worse than screening noisily.
  • OFAC data is public domain (U.S. Treasury). UK FCDO data is reused under the Open Government Licence v3.0. We publish our matching method; we never claim certainty we don’t have.

FAQ

Straight answers.

How do you handle false positives?

Common names produce low-confidence fuzzy matches — that’s expected. In our calibration on the live OFAC file, no common-name query (JOHN SMITH, MARIA GARCIA, LI WEI, and others) ever reached medium confidence, let alone high. JOHN SMITH returned zero matches. Low-confidence matches are always shown, never hidden, so you can see the noise and dismiss it. A date of birth or ID number is the fastest way to turn a fuzzy name match into a reliable answer.

Do you cover EU and UN sanctions lists?

At launch we screen the OFAC SDN list (U.S., public domain) and the UK FCDO consolidated list. The EU consolidated list is added once a free EU Login account is in place for its daily download. The UN list is excluded: its terms block commercial use, and EU regulations implement UN designations, so coverage largely overlaps via the EU list.

What does an AI agent need to call SanctionScreen?

An API key and the two tool definitions — nothing else. The tools ship with few-shot input/output examples in their descriptions, they’re available over MCP (no extra fee), and results are structured JSON your agent can parse directly. Idempotency keys let an agent retry safely without double-billing. Machine-readable docs: llms.txt.

What are the free tier limits?

50 calls/month total across both tools. No card required, no burst allowance, and a hard cap — abuse can’t run up a bill on your key, and a stuck agent can’t burn quota for free (retry traffic is quota-counted too).

Do I need a card for the free tier?

No. Free is free: 50 calls/mo, no card, no expiry gimmicks.

Get a free API key

SanctionScreen is in private beta. API keys are issued manually right now — 50 calls/month, no card, no expiry gimmicks. For agents, the MCP endpoint is already live:

https://api.sanctionsscreen.io/mcp

Integrating today? Start with the agent docs (llms.txt) — everything an agent needs to self-serve.