The fraud rule that covers 99.8% of claims, and splits the bill.

Three design choices in this regime are worth understanding even if you are not a payment firm, because each of them is a decision about where to put an incentive.

The cap was set by coverage, not by affordability. £85,000 is not a round number chosen for neatness; it was justified by what share of claims it reimburses in full. Publishing both the volume figure and the value figure is unusually honest, because together they say plainly that the 0.2% of claims above the cap carry 10% of the losses.

The clock is five business days. A protection that pays eventually is a different product from one that pays this week, and the deadline is what turns it from a right into an experience.

The split is 50:50. This is the interesting one. The firm that sent the money and the firm that received it each pay half.

Why the split is the real mechanism

Before this, the incentive to stop a scam sat almost entirely with the sending bank, which is the party least able to see what happens next. The receiving firm holds the account the money lands in, the pattern of accounts opened, and the behaviour after receipt — and had limited reason to act.

Making the receiving firm pay half moves the cost to where the information is. That is a general principle worth borrowing well outside payments: put the liability where the evidence lives, or the party who could have prevented the loss never finds out it happened.

What it means for an adviser who is not a payment firm

Two things, and neither is a compliance obligation.

First, your clients now have a protection they probably do not know about, with a cap high enough to cover almost any loss they will suffer. A client who has been defrauded and has given up is worth asking about it, and the five-day expectation is worth telling them.

Second, this protection covers Faster Payments and CHAPS scams. It does not cover investment losses, a bad recommendation, or a transfer the client made deliberately on advice — and the gap between "I was tricked into paying" and "I was badly advised" is exactly where a complaint about you begins rather than about a bank. That boundary is worth being precise about in writing.

What this does not tell you

The figures are the PSR's own, published when it confirmed the cap, and they describe claims as they were then. The mix moves and the regime has been argued over since; treat 99.8% and 90% as the justification given for the level rather than as a current statistic.

What has happened since is measured too. An independent evaluation the PSR published on 1 July 2026 estimates that APP losses on Faster Payments fell by about 21% — £73 million a year — after the requirement began, and that the reimbursement rate rose from 54% to 65%. In the first quarter of 2026, 82% of claims were resolved within five days.

UK Finance's total for 2025, which counts every APP loss its members recorded, still rose, to £576.4 million. The two measure different things — an estimated effect of one rule, and a total — and we would not set one against the other. The PSR says it will consult on improvements before the end of 2026.

It also does not describe the consumer standard of caution or the exceptions, which decide the cases people actually disagree about. And the wider redress framework around all of this is being rebuilt — that is the CP26/9 note, and what the Ombudsman is actually receiving is in the complaints data. Our work with regulated firms is about the record either side of advice, which is where this boundary gets decided.