Treasury Sanctions Law Firm and Two Ship Agents Behind Iran's Shadow Fleet Paper

The Analysis

Two dates matter here, and confusing them is how people get this wrong. The Treasury bulletin was distributed on 21 October 2026; the press release carries 22 October 2026. The deadline that binds behaviour is a third one: transactions involving designated persons have to be wound down by 20 November 2026, US Eastern time, under two general licences issued with the action. The authority is Executive Order 13902.

Operation Economic Outcast designated Karnakis and Karnakis, a law firm based in Panama and Greece, together with five individuals connected with it, on the basis that the firm registered shell companies for shadow-fleet vessels. It designated Amigoz, a Singapore ship agency, and Whales Shipping Line in Dubai, plus vessels. FinCEN is separately using a Bank Secrecy Act subpoena against a Greek bank. Chinese reporting puts the vessel count at nine and repeats the 20 November date; I could not verify the nine against the primary bulletin, so I am treating it as reported rather than confirmed.

What is new is not the hull count. A law firm and two ship agents are now inside the designation perimeter, and that moves the check off the ship and onto the paper behind the ship.

Start with the rule, stated precisely, because everything else rests on it. The instrument doing most of the work is not the designation itself. It is the 50% rule that sits behind every OFAC designation, and people quote it loosely, so let me quote it tight: under the standing OFAC position, an entity owned 50% or more, directly or indirectly and in the aggregate, by one or more blocked persons is itself blocked, whether or not it appears on the Specially Designated Nationals list under its own name. Strictly speaking, being left off the list buys an entity nothing. The precondition for the rule to apply is an ownership fact — who holds what, through how many intermediate companies, in what proportion. It is not a listing fact.

That distinction is the entire reason this action is heavier than a vessel addition. Karnakis and Karnakis is a law firm, Amigoz is a Singapore ship agency, Whales Shipping Line is a Dubai shipping line, and none of them is a hull. When a chartering desk or a cargo interest runs the usual check against a vessel list, none of these names surfaces, because none of them sits in the field the check reads. The vessel is the field that gets screened. The firm that registered the shell company behind the vessel is not a field at all.

Why now, and who is pushing. Designating ships has a structural limit: a hull can be renamed, reflagged, sold into a fresh single-purpose company and put back to sea with a new paper trail, and the next round of designations has to start again from the new name. The enabler layer does not behave like that. A registry entry, an agency appointment, a corporate filing, a bank mandate — these are durable records held by institutions that keep files, and one designation of the enabler reaches every vessel that ever used it. On that reading, the choice of targets looks like a response to diminishing returns on hull designations rather than a change of policy.

Treat that as my inference, not as something Treasury said. What Treasury did say is that FinCEN is using a Bank Secrecy Act subpoena against a Greek bank. A subpoena is a document request, and a bank is where the ownership and control facts the 50% rule turns on actually live. Put the designation of a law firm next to a subpoena to a bank and both point at the same place: the evidence here is documentary, and the documents are held by intermediaries rather than by shipowners.

The timing is tighter than the calendar suggests. The press release is dated 22 October and the wind-down deadline is 20 November, US Eastern time — 29 days, counted in Eastern time, which bites anyone releasing payments out of Asia or the Gulf. A transfer initiated on 20 November local time in Singapore can land on 19 or 20 November Eastern depending on the hour; one initiated on 21 November local time is already outside. There is also a lag people forget: the designation is effective on publication and the general licences are effective on publication, but your screening vendor's data is effective whenever the vendor pushes it — commonly inside a day for the list itself, materially later for the 50%-owned derivatives.

On who is touched, sorting matters, because "shipping is affected" is not an answer. Layer one is the named parties plus every entity 50% or more owned by them: for those, the general licences are the only route to doing anything at all, and the route closes on 20 November. Layer two is counterparties with open business touching them — a charterer with a running fixture, a cargo interest with cargo laden or booked, a bunker supplier holding an unpaid invoice, a port agent holding funds. Theirs is a deadline problem rather than a prohibition problem, and it is measured in days.

Layer three is banks and insurers; FinCEN's subpoena is aimed at a bank, and correspondents respond by questioning their own customers on their own timetable, which nobody outside the bank controls. Layer four is registries, flag administrations and classification societies, where the consequence is regulatory and reputational rather than sanctions exposure in the strict sense. There is one more layer, the largest and least visible: everyone who will later have to prove they checked.

On size, I am not going to invent a number. The named group is small and countable — one firm, five individuals, two agencies, an unconfirmed vessel count. The 50% rule multiplies it by an unknown factor. If each named entity sits above three majority-owned affiliates that were never separately listed, your screening file triples; if the ownership trees are flat, it barely moves. Nobody has published the trees. Anyone handing you a confident population figure for this action is guessing. The honest posture is to size your own file bottom-up from your own counterparty list rather than top-down from a headline.

The transmission sequence I would plan against runs like this. Publication day: new business with designated persons stops, and existing business continues only inside the general licences. Within two to three days: list data reaches screening systems and the first false positives appear, mostly name similarity rather than true matches. Days three to ten: the 50%-derived data starts arriving from vendors, and this is where real hits tend to sit, because the derivative pass is what catches an unlisted affiliate. Days ten to twenty-nine: the wind-down itself, including cargo already afloat, demurrage accruing on a vessel nobody wants to discharge, and payments scheduled before 22 October that are now awkward. After 20 November Eastern: the authorization ends and anything unfinished becomes a different kind of problem. The second wave — bank questionnaires, counterparty representations — usually starts after the deadline rather than before it, and it is the most staff-intensive part.

Here is the part the coverage did not say, and it is the part that will cost people money. The perimeter moved from the hull to the paperwork layer. Once a law firm and two ship agents are designated, the counterparty check has to run one level deeper than the vessel: who registered the shell company, who acts as agent for the vessel, who holds the bank mandate. A screening process built around IMO numbers and vessel names will now return clean on a transaction that is not clean, and it will return clean with a timestamp. That is worse than returning nothing, because the timestamp looks like evidence.

The trap is how the two instruments interact, and there is an exception buried in it that gets misread. The general licences authorise the wind-down of transactions involving designated persons until 20 November. That authorisation answers one question: may I transact. It does not answer the other one, whether the counterparty on the other side is clean. So you have 29 days in which doing business is permitted and in which every payment still needs a screening trail with a defensible basis. And the old answer, "the ship is not on the list", stops working inside that window: the ship can be absent from the list while the agent behind it is present, and the 50% rule can reach the agent's affiliates without any of them being named. Don't rush to a conclusion on the strength of a clean vessel check.

On the paperwork side, Derek Xu puts it more shortly than I can: the file you build during the window is the file somebody asks you for afterwards, so the record has to be built at the moment of the decision, not reconstructed when the question arrives. I would add one thing from my side. The record should say what was screened, against which list version, at what time, and what the decision rested on. A screening result without a list version and a timestamp is not a record; it is a screenshot.

Now the conditions under which this flips, and there are four. One is the scope of the two general licences. A general licence is a text with defined categories, and the categories are what they say, not what a summary says. If the authorisations are broad — covering, say, safe docking and discharge of cargo already laden — the pressure in the window is operational rather than legal. If they are narrow, transactions outside them need a specific licence, and a specific licence is not a form you fill in during a wind-down. I have not seen the licence text reproduced in the coverage I have read, so the honest instruction is to read both licences before you plan, and not to plan on the assumption that "wind-down" is one single category.

Another is the burden of proof under the 50% rule. Ownership is a fact about the world; your screening result is a fact about your data. Panamanian and Greek corporate records do not always disclose beneficial owners, and where they do not, a clean screen tells you that you found nothing, not that nothing is there. That gap is where the conclusion can turn. If a counterparty's ownership chain is opaque, the defensible move is to ask for a written ownership representation and to record that you asked, rather than to treat silence as clearance.

The position of non-US persons is where both directions get overstated. US sanctions reach non-US persons through a nexus: a US person in the chain, US-dollar clearing through a US correspondent, US-origin goods, software or technology, or conduct an authority treats as causing a US person to act. A non-US trading house clearing euros through non-US banks with no US persons involved has a materially different exposure profile from one clearing dollars through New York, and the gap is wide enough that a single global policy is usually wrong at one end or the other. The mirrored error is assuming that sitting outside the US perimeter removes the question. It does not, because correspondent banks apply their own standards, and a bank's standard is a commercial fact you cannot argue with. Strictly speaking, your bank's risk appetite is not the law, but it decides whether your payment moves.

The EU and the UK are not on the same clock, and that is the fourth condition. Their designations are separate lists with separate criteria and separate amendment dates; an entity can be US-designated while remaining undesignated in the EU and the UK, or the reverse. For an operation centred on Rotterdam or London this is a three-list problem, and each list needs its own check with its own timestamp. Where the lists diverge, EU operators also have to think about the EU blocking statute, a different instrument with a different purpose that cuts against automatic compliance with a US designation. That tension is one for counsel. From a screening practice standpoint the operational point is short: you cannot screen one list and call it three.

Let me put a number on the workload, with every assumption written out, because a screening policy you cannot staff is not a policy. Assume a mid-size commodity desk: 400 payments a month, average ticket USD 850,000, monthly throughput around USD 340 million. Assume each payment touches five nodes capable of carrying designation risk: the vessel and its IMO number, the registered owner on the registry, the commercial or shipping agent, the paying or receiving bank, and the contracting legal entity on the invoice. Assume the existing process screens two of those five, vessel and contracting entity, which is typical when a document is checked against a list. Assume the observed flag rate on the covered layers is 1.5% of payments. Assume each of the three newly covered nodes adds an independent 2% chance of a flag.

On those assumptions the baseline is 0.015 x 400, roughly six flagged payments a month. Adding three nodes at 2% each gives an incremental probability of 1 - 0.98 cubed, which is 5.88%; combined with the baseline, 1 - (0.985 x 0.9412) is 7.27%, or about 29 flagged payments a month. The delta is 23 extra escalations. At 45 minutes of triage each — pulling corporate records, walking the ownership chain, writing the escalation note — that is 17.25 hours a month, roughly two and a half working days. At a loaded internal analyst cost of USD 70 an hour, the recurring bill is about USD 1,200 a month, or USD 14,500 a year, before any external counsel.

The other side of the same arithmetic is exposure. Assume that one in eight escalations turns out to involve a genuine ownership link to a designated person — an assumption, and a generous one, since nobody has published a base rate. That is roughly four payments a month at USD 850,000 each, about USD 3.4 million of flows that need a documented basis for being made or being stopped. Run the same model against the wind-down window instead of a month: 29 days, proportionally about sixteen or seventeen hours of triage inside the window, on a desk that also has cargo to discharge and contracts to close. Two things should be plain from this. The cost is small relative to the flows. The staffing is not small relative to the window, and seventeen hours in 29 days is exactly the difference between a process that runs and a process that gets skipped on the busy days.

What I would do with the days left, framed as screening and disclosure practice rather than as legal advice, because the legal characterisation belongs to counsel. Re-run the whole counterparty file against the current list plus a 50% ownership pass, and archive the list version and the run timestamp alongside the result. Draw the five-node map for every open contract: vessel and IMO, registered owner, agent, bank, contracting entity. Write a one-page decision note for each open transaction that touches a designated person, saying which general licence it sits under, which sentence of that licence, and who signed off — and send the ones you are unsure about to counsel rather than settling them at the desk.

Give every payment in the window a screening record naming what was checked, the list version, the time, the conclusion and the person who decided. Ask your banks in writing what ownership standard they apply and how often their ownership data is refreshed, because their answer tells you whether your payment clears. Put a screening representation and a suspension right into new charters and agency appointments, with the trigger written as "any party, or any holder of 50% or more of any party, becomes designated".

On alternatives: reflagging a vessel does not touch an agency designation, switching the paying bank does not touch an ownership link, and moving the contracting party to an affiliate can make things worse if that affiliate is majority-owned by the same person. The tempting move, freezing everything that looks remotely connected, carries its own bill — legitimate cargo sits, demurrage runs, a customer relationship goes. The traps I would watch are treating a general licence as a general safe harbour, relying on a vendor's derivative data without knowing its refresh lag, reading a clean vessel screen as meaning anything, and rebuilding the file after the deadline instead of during the window. There is one quiet trap as well: if you find a match on 19 November you have a day to decide, and the decision will be made by whoever happens to be in the office, so write the escalation rule down before you need it.

Where does this not hold. If your operations have no US nexus and no dollar clearing, what I have described is your bank's constraint more than yours, and the binding standard may be the EU or UK list instead. If the two general licences turn out to be broader than the summaries imply, 20 November is an administrative milestone rather than a cliff. If a counterparty's ownership is fully disclosed in a registry that actually publishes beneficial owners, the 50% rule becomes checkable rather than arguable and the screening conclusion is correspondingly stronger. If the vessel count differs from the nine carried in Chinese coverage, the size of the file changes and nothing in the method does. And if you are reading this after 20 November, the wind-down question is closed and what remains is the record: what you knew, when you knew it, and what you did with it. That is the version of this that stays.

↑ Back to the key points

— By Dr. Ingrid Voss