· Yair Knijn
A one-time check on a multi-year policy: the assumption that aged into a violation
The Underwriting Director signs off on a five-year construction liability program. The screening report sits in the file, dated the day of bind, every party clean. To the director, that report is the compliance artifact for the whole policy. The check happened. The box is ticked. The term runs to 2031.
That is the trap, and it is a quiet one. The director has silently equated the lifespan of one screening decision with the lifespan of the contract. A decision that was true for about a day is being asked to stand for sixty months.
The mismatch between a point-in-time screen and a multi-year exposure
A screen tells you the state of the world at the moment you ran it. A sanctions list is a database with a changelog, not a fact about a person. OFAC updates the SDN list on no fixed schedule, sometimes several times a week, and a name added on a Tuesday is enforceable against you on that Tuesday whether or not you looked. Your policy, meanwhile, does nothing. It just sits there accruing exposure against a screen whose evidentiary value decayed almost immediately.
OFAC made this explicit in its updated insurance FAQs. Screening only before policy issuance, it says, is critical but "would not likely achieve the desired level of compliance," because the government may designate an existing policyholder or named beneficiary after you bound the risk. The agency lists the moments that should re-trigger a screen: renewal, amendment, claim submission, claim payment, and any update to the sanctions lists themselves. A one-time check at inception is the one pattern OFAC singles out as insufficient.
Everything that changes mid-term: lists, ownership, PEP status, beneficiaries
The list is only the most obvious moving part. Across a five-year term, the parties themselves drift. Consider what actually shifts under a static policy number:
- The UBO changes. A holding company restructures, a stake is sold, and the beneficial owner who was clean at bind is now someone you never screened.
- PEP status turns on. A board member becomes a deputy minister. A founder's brother is appointed to a state enterprise. None of this touches the policy document.
- A beneficiary is added or assigned. The named loss payee on year four was not in the file you screened in year one.
- The list catches up to reality. A designation that was pending at bind lands eighteen months later, against a party already on your book.
Why "we screened them" becomes "we screened them once, years ago"
When the regulator opens the file, the sentence the director is relying on does not survive contact. "We screened all parties at onboarding" sounds like a control. Written out fully, it reads "we screened them once, at bind, and never again across the life of the contract," and that is not a control, it is a gap with a date on it. The enforcement risk is not abstract. OFAC's own guidance ties inadequate screening frequency directly to enforcement exposure: if a violation occurs in the silent years, the fact that you screened at issuance does not mitigate it, it documents exactly when you stopped looking.
The deeper problem is that the director never priced the obligation as continuous. It was booked as a transaction cost at inception, one line item, paid once. The actual cost is a recurring liability that runs the full term, and it was put on the books as if it were settled at bind.
Treating the policy term as a continuous screening obligation
The fix is a reframing, not a bigger one-time effort. Coverage is live from bind to expiry, so screening has to be live across the same window. Every party on every in-force policy gets re-evaluated whenever the list moves or the party's profile changes, and each clearance carries its own short shelf life rather than inheriting the policy's. The unit of compliance is the policy-day, not the policy.
This is the model InsureGuardAI is built around. Each tenant gets a workspace where in-force policies stay continuously matched against list updates, ownership and PEP changes surface as events rather than being discovered at claim time, and every clearance keeps its raw match response so the file shows not just that you screened, but when, against which list version, and why a match was cleared. The question stops being "did we screen them" and becomes "are they still clear today," which is the only version of the question the term actually asks. See how continuous screening works.