The month my bank was allowed not to read the name, at 82K MRR

That was the second thing my bank's funds transfer team said to me. The first was that the money had gone. We were at 82K MRR, I had paid a supplier invoice of 18,400 dollars, and I had transposed two digits of an account number while copying it out of a renewal email. The name on the payment order was my supplier's, spelled correctly. The number belonged to somebody neither of us had heard of. I had assumed, the way you assume the floor is there, that a bank receiving a name and a number that disagree would stop and ask. It took three weeks to learn that the law has considered this carefully and decided the opposite, and that the one thing which would have saved me was a document I had already signed without reading.

Quick answer (2026)

This is a composite founder diary. The company and the numbers are invented; the law is real, quoted from primary sources, and current as of October 2026. If you send a payment order where the beneficiary name and the account number point at different people, Section 4A-207 of the Uniform Commercial Code lets the receiving bank pay on the number and says in terms that it "need not determine whether the name and number refer to the same person". You may still escape the obligation to pay, but only if you are not a bank, only if you prove the stranger was not entitled to your money, and only if your own bank cannot prove you were given notice that numbers govern. That notice is routinely buried in the funds transfer agreement you signed at account opening. And if the escape works, the right to chase the stranger is not yours.

The payment order I actually sent

A payment order is not a request. It is an instruction with fields, and the fields do not carry equal weight. Mine carried a correct name, a wrong number, an amount and a date. The beneficiary's bank, which had never heard of me and owed me nothing, received an instruction naming one person and numbering another.

I expected a bounce. I got a deposit, because the statute anticipates this disagreement and resolves it. Subsection (b) applies where a payment order "identifies the beneficiary both by name and by an identifying or bank account number and the name and number identify different persons". It then gives the receiving bank a rule rather than a duty of inquiry. The money moved on a Tuesday. I found out on a Thursday, when the supplier chased the invoice.

The sentence that decided it

Here is the operative language, and it is shorter than you would like.

Section 4A-207(b)(1), verbatim: "Except as otherwise provided in subsection (c), if the beneficiary's bank does not know that the name and number refer to different persons, it may rely on the number as the proper identification of the beneficiary of the order. The beneficiary's bank need not determine whether the name and number refer to the same person."

Two phrases do the work. "Does not know" is a knowledge test, not a carelessness test. "Need not determine" forecloses the argument I had already drafted in my head, that a competent bank would have checked.

This is not theoretical. In a 2025 decision the Fourth Circuit reversed a judgment of 558,868.71 dollars that a trial court had awarded against a credit union on exactly that reasoning, holding, in the words of a client alert on the decision, that "Because there was no evidence of actual knowledge presented in the case, it was error for the court to have held 1st Advantage liable on a finding of negligence or commercial unreasonableness." The bank in that case ran an automated system that generated name mismatch warnings and nobody read them. That was still not knowledge. The court's reason was volume: "Countless discrepancies can arise inadvertently and harmlessly."

So the route I assumed existed, suing the bank that took the money, is narrow by design.

The escape I did have, and the thing that deleted it

Almost nobody writes about the next subsection, and it decides whether you personally are out the money. Subsection (c) governs what happens between you and your own bank once the beneficiary's bank has lawfully paid on the number.

It splits into two populations, and for once the small operator is on the better side. If the originator is a bank, the text is a flat sentence with no relief in it: "If the originator is a bank, the originator is obliged to pay its order." If the originator is not a bank, there is a door. You are not obliged to pay, provided you prove the person identified by number was not entitled to receive payment from you.

Then the door closes, on a condition that has nothing to do with your conduct. You are off the hook "unless the originator's bank proves that the originator, before acceptance of the originator's order, had notice" that payment might be made on the number even where it identifies someone other than the named beneficiary. And the statute tells your bank precisely how to discharge that burden: "The originator's bank satisfies the burden of proof if it proves that the originator, before the payment order was accepted, signed a writing stating the information to which the notice relates."

One signature. Not negligence, not anything I did on the day. A clause I accepted at account opening, saying in substance that numbers govern. I found it, in the funds transfer terms, in the same block as the cut off times.

One word, changed in 2024

Article 4A is not federal law. It is a uniform text each state enacts, so the sentence deciding your case is the one your state passed. I read mine against the enacted version in the jurisdiction where that Fourth Circuit case arose, and the words are not identical.

The uniform text as published says the bank wins by proving you "signed a writing". Virginia's enacted section says it wins by proving you "signed a record". Its history line reads 1990, c. 9; 2024, c. 652, so I pulled the Act. Chapter 652 of the 2024 session amends and reenacts section 8.4A-207 among dozens of others, and prints the change inline, strike and insert, as "signed a writing record".

That is not cosmetic, and the code defines both words. A writing "includes printing, typewriting, or any other intentional reduction to tangible form". A record "means information that is inscribed on a tangible medium or that is stored in an electronic or other medium and is retrievable in perceivable form". Tangible form, or tangible form plus electronic. The class of thing that satisfies your bank's burden got larger, by one word, in 2024.

I am not claiming this flipped any particular case. I am claiming the single input deciding whether you owe the money is now easier for your bank to establish than the uniform text suggests, and that you will not learn it from the uniform text.

The other door was already shut

A second provision looks like it was written for me, and I spent two days being wrong about it. Section 4A-205, the erroneous payment order rule, covers precisely my situation: an order that "erroneously instructed payment to a beneficiary not intended by the sender". Duplicate payments and overpayments are in there too.

But read the opening condition. It applies only "If an accepted payment order was transmitted pursuant to a security procedure for the detection of error". And the definition of a security procedure is disjunctive. It is a procedure agreed between customer and bank "for the purpose of (i) verifying that a payment order or communication amending or cancelling a payment order is that of the customer, or (ii) detecting error in the transmission or the content of the payment order or communication".

Purpose (i) is authentication: proving the order is really from you. Purpose (ii) is error detection: catching that the order is wrong. Almost every business has the first. I had the first. They are not the same procedure, and 4A-205 is keyed to the second. A strong authentication setup bought me nothing here, which is the single most useful thing I learned.

Even where it applies, the relief is not free. You must prove both that you complied with the procedure and "that the error would have been detected if the receiving bank had also complied". And it comes with a duty running the other way: once notified, "the sender has a duty to exercise ordinary care, on the basis of information available to the sender, to discover the error", and to tell the bank "within a reasonable time, not exceeding 90 days".

The chase follows the loss

The last subsection is four lines long and I would have got it wrong in a negotiation.

Having spent a week arguing I was not obliged to pay, I had assumed winning meant two good things at once: I keep my money, and I still pursue the stranger whose account swallowed it. Subsection (d) says otherwise. Where the beneficiary's bank rightfully paid on the number, the money is recoverable from the recipient under the law of mistake and restitution, and the statute then allocates that right. If you are obliged to pay, you have the right to recover. But "If the originator is not a bank and is not obliged to pay its payment order, the originator's bank has the right to recover."

The chase belongs to whoever bears the loss. That is coherent, and nobody had told me. It also means the two outcomes are not a spectrum. They are a switch.

What I got wrong

I was wrong three times, and recording them is the point.

I assumed the absence of a carelessness standard meant the whole area was vague and therefore arguable. It is not vague, it is allocated. Within 4A-207 and 4A-205 the words willful, negligent, reckless, good faith and safe harbor do not appear at all. The only knowledge test in 4A-207 is the beneficiary bank's ignorance, which protects the bank, and the only standard of care in 4A-205 is ordinary care, which is a duty I owe. Uncertainty was not sitting on my side of the table.

I assumed being a small company was irrelevant to a commercial statute. Here it is the only thing that opened the door at all, because subsection (c) gives relief to a non bank originator and gives a bank originator none.

And I assumed, when I found that Virginia said "record" where the uniform text said "writing", that I had found a drafting slip. I had found a dated amendment, which is a different and much more useful object.

What actually happened

Limits

This is one composite narrative, not advice, and three gaps are worth naming. I read the uniform text as published by Cornell and the enacted text of one state; I did not read the current promulgated text from the uniform law commission itself, so I can say the two sources I read differ by that word and not who moved first. I have not researched whether an electronic account agreement already counted as a writing in Virginia before 2024 under general electronic signature law, so the practical size of that change is unestablished. And the appellate decision is reported here through a law firm's client alert, read in a browser because the site refuses automated fetchers; I did not obtain the slip opinion.

The one thing I would tell you

Go and open your funds transfer or treasury agreement this week, search it for the word "number", and find out whether you have already been given notice that the number governs. That single clause is the difference between a wrong digit being an expensive inconvenience and a wrong digit being final, and it is sitting in a PDF you can read in four minutes.

Then ask one follow up question, which is the one I did not know to ask: is the security procedure we have with you for verifying that orders are ours, or for detecting errors in them? If the answer is the first, and it usually is, you should know that the provision written for your exact mistake does not reach you.