let’s go back to Scenario A: there has to be, in order to effectuate the instruction, an account relationship of some kind, doesn’t there? So why is that case different? You’ve got assets being allocated for a sanctioned party – the only difference is that, in Scenario A, it’s being taken from an existing pooled asset. Or am I missing something?

You’re not missing something, and it’s a sharp catch — there genuinely is an “account relationship” and something IS being drawn from a pooled asset. But the question is whether that operational fact amounts to a legal interest for the designated person, and here payment law itself already draws the line you’re looking for, independent of sanctions law entirely.

A wire transfer isn’t one asset moving through a pipe — it’s a chain of separate obligations

Under the commercial law that actually governs funds transfers (UCC Article 4A in the US; the EU’s Settlement Finality Directive and the payment-services framework in Europe), a beneficiary acquires zero legal claim against their own bank until that bank “accepts” the payment order — the beneficiary’s bank is only obliged to pay the beneficiary once it accepts the order, and acceptance itself is defined narrowly: the earliest of the bank paying the beneficiary, notifying them of receipt, or receiving settlement for the order. Before that moment, nothing has legally reached the beneficiary — the money belongs to the recipient, free of any strings, only once finality is reached. This isn’t a sanctions-specific fiction. It’s the ordinary rule that governs every wire transfer on earth, sanctions or not — it’s what determines, say, whose money it is if a bank fails mid-transfer, or if a payment gets mis-keyed to the wrong account.

So what is actually happening at your end when you debit Customer A?

Bank A debiting Customer A’s account and sending a payment message isn’t Bank A moving the designated person’s property — it’s Bank A discharging (or beginning to discharge) its own contractual obligation to Customer A, funded out of Bank A’s own balance sheet and correspondent liquidity. That pooled asset is Bank A’s, being drawn down pursuant to A’s instruction; the designated person is a total stranger to that relationship until their own bank actually accepts the order on their behalf. If you refuse at that point, nothing reverses because nothing left — Customer A’s account was never debited (or is simply uncredited if it briefly was), and there’s no moment where the designated person held anything that now has to be “returned.” That’s precisely why refusal doesn’t require a licence the way unwinding a completed freeze does: there’s nothing to unwind.

Why “belonging to/held/controlled by” tracks this, and why OFAC doesn’t

The EU-family test (belonging to, owned, held, or controlled by) is simply importing the ordinary payment-law vesting point rather than inventing its own. OFAC’s blocking regime is the outlier specifically because its “interest of any nature whatsoever, direct or indirect” definition is deliberately written to reach behind that ordinary vesting rule — US commercial law (the same UCC 4A) agrees with the EU-family approach that the beneficiary has no claim pre-acceptance, but OFAC’s sanctions regulations override that for blocking purposes, treating mere intended-beneficiary status as sufficient. So the divergence being described here isn’t a difference in what property law says happens to a payment in flight — both systems’ underlying commercial law agrees on that. It’s that only one of them layers a sanctions-specific rule on top that disregards it.

This also explains the countermand test from the UK “trapped payments” guidance covered previously — asking whether a designated bank still had the right to recall the funds is just asking whether finality/irrevocability had occurred yet. Same doctrinal line, applied to figure out whether the freeze duty travels with the money through a correspondent chain. And it’s why an intermediary bank in the EU-family model typically has nothing of the designated person’s to freeze mid-chain at all: under this same logic, none of the intervening correspondent legs constitute the beneficiary holding anything either, right up until the final beneficiary’s bank accepts.

One thing worth keeping separate: this settles whether the Scenario A/B line is legally real rather than cosmetic — it doesn’t reopen the policy question of whether refuse is a strong enough response on its own. Those are different questions.