A business receives an invoice from a supplier it has worked with for years. The amount is correct, the correspondence looks familiar and the request to update the supplier’s bank details seems entirely reasonable. The payment is approved through the normal process and sent to the new account.
Unfortunately, the instructions came from a criminal impersonating the supplier, and the money has gone elsewhere.
It’s a familiar example of payment diversion fraud, and one that highlights a difficult problem for financial institutions. The person making the payment may have followed every required authorisation step, while the decision itself was based on fraudulent information.
In September, the City of London Police reported that victims had lost more than £101 million to payment diversion fraud during the 2025/26 financial year. Although reported losses had fallen compared with the previous year, the number of reports increased by 3% to 3,657.
Criminals are exploiting established business relationships, compromising email accounts and creating convincing messages to persuade people to send legitimate payments to fraudulent destinations.
Our Key Business Concerns for 2026 report identified payments fraud as one of the areas deserving particular attention.
As we continue our November look back at three of those concerns, we’re focusing on authorised payment fraud and the challenges it presents for institutions trying to identify deception before money leaves an account.
The fraud often begins long before the payment
The latest figures from UK Finance help illustrate the scale of the problem.
Its Annual Fraud Report, published in June, found that authorised push payment (APP) fraud losses increased by 19% in 2025, reaching £576.4 million. The number of cases also rose by 7% to 248,070.
Purchase scams accounted for 71% of cases, but investment fraud was responsible for the greatest financial losses, reaching £221.5 million, an increase of 40% on the previous year.
These involve very different circumstances.
Someone purchasing a product through an online marketplace may be persuaded to transfer money directly to a seller who doesn’t exist. An investment scam may develop over weeks or months, with repeated conversations, apparently credible documentation and a relationship built around promises of financial returns.
By the time a payment is made, the criminal may have spent considerable effort making the transaction appear legitimate.
UK Finance also found that 66% of APP fraud cases originated online, while another 17% began through telecommunications.
Much of the activity leading to a fraudulent payment therefore takes place outside the financial institution’s own systems.
The bank can see the payment instruction, the customer’s account activity and the information available about the receiving account. It may have little visibility of the conversations, advertisements or false claims that persuaded the customer to transfer their money.
This creates a gap between what the customer has experienced and what the financial institution can reasonably establish about the transaction.
When convincing someone becomes the attack
Recent police warnings show how the methods used to deceive customers are developing.
In October, the City of London Police reported that more than £24.3 million had been stolen through AI-enabled investment fraud between January and August 2026, with over 1,300 reports received during that period.
These figures concern AI-enabled investment fraud more broadly, rather than APP fraud alone, but the methods described are particularly relevant.
Criminals used AI-generated images, videos and false endorsements to make investment opportunities appear credible. Victims were then contacted by people posing as financial advisers or account managers and encouraged to transfer money.
Some were shown supposed profits through professional-looking trading platforms before being persuaded to invest larger sums. When they attempted to withdraw their money, they encountered further demands for fees or charges.
The use of AI is noteworthy, but the wider issue is how much deception can take place before the financial institution becomes involved.
Strong customer authentication remains an important protection against unauthorised transactions. Confirmation of Payee and transaction monitoring can also help identify potential fraud.
However, authentication establishes that the person making the payment has met the required verification steps. It cannot establish whether that person has been deceived about the transaction’s purpose.
A customer may use their normal device, log in from their usual location and successfully complete every authentication requirement, yet still be acting on instructions from a fraudster.
That presents a difficult challenge for payment providers. How do they recognise a fraudulent decision when the payment instruction itself appears genuine?
Behavioural monitoring and transaction analysis can help identify unusual activity, but neither provides a complete picture.
A legitimate transaction can look unusual, while a fraudulent payment may be consistent with the customer’s established behaviour.
Firms therefore have to make decisions with incomplete information, balancing the need to prevent fraud against the disruption caused by unnecessarily challenging genuine transactions.
Reimbursement is improving, but prevention remains difficult
There have been important developments in the way financial institutions respond to APP fraud.
Since October 2024, mandatory reimbursement requirements have provided greater protection for eligible victims of APP scams involving UK Faster Payments and CHAPS transactions.
In July, the Payment Systems Regulator reported that 88% of the value of reimbursable APP scam claims covered by its Faster Payments rules had been returned to victims during the first 18 months of the policy. That represented approximately £316 million.
An independent evaluation by Frontier Economics, also published in July, estimated that APP fraud losses sent over Faster Payments had fallen by around 21% following the introduction of the reimbursement requirement, equivalent to approximately £73 million a year.
The evaluation also found that the policy had strengthened incentives for payment service providers to invest in fraud prevention.
These are encouraging findings, particularly as they suggest that the reimbursement framework is influencing behaviour as well as improving outcomes for victims.
They also need to be considered alongside UK Finance’s reported increase in APP fraud losses during 2025.
The apparent difference is partly explained by what the two studies measure.
UK Finance reports a broader range of APP fraud, while Frontier Economics examined the effects of specific policies on Faster Payments using a different methodology and measurement period.
The figures are therefore not directly comparable.
What emerges is a complicated picture. Reimbursement protection has improved, and there is evidence that the new rules are encouraging stronger fraud prevention. At the same time, the broader payments environment continues to experience substantial losses.
For financial institutions, this raises questions about how success is measured.
Reimbursing a victim reduces the financial harm they suffer, but it doesn’t necessarily prevent the criminal from receiving the money. A reduction in losses within a particular payment system is encouraging, although it doesn’t establish what is happening across every other route through which fraudsters can move funds.
Both prevention and reimbursement matter, but they tell us different things about the effectiveness of the overall response.
The limits of what an individual bank can see
One of the more difficult aspects of APP fraud is that the information needed to recognise it may be spread across several organisations.
The sending bank knows its customer’s account history and payment behaviour. The receiving institution may have information about the destination account, previous incoming payments and any activity suggesting that the account is being used to receive fraudulent funds.
An online platform or telecommunications provider may hold information about the initial contact between the criminal and the victim.
Each organisation sees part of the activity, but none necessarily has the complete picture.
This is one reason the responsibilities of receiving institutions have become more significant.
Under the UK’s APP fraud reimbursement arrangements, sending and receiving payment service providers generally share the cost of reimbursing eligible victims equally.
That creates a financial incentive for both sides to identify and prevent fraud, including the misuse of accounts receiving stolen money.
The difficulty is determining whether the available information is sufficient to justify intervention.
An unfamiliar beneficiary or a sudden change in payment behaviour may indicate a problem. Equally, customers routinely make legitimate payments that fall outside their normal patterns.
A transaction that looks entirely ordinary could also be part of a carefully constructed scam.
For risk and compliance teams, the challenge extends beyond the performance of individual transaction-monitoring controls.
They need to understand where useful warning signs might arise, whether those signals reach the right people and how consistently decisions are made when something looks suspicious.
A fraud prevention framework may contain several well-established controls, but weaknesses in the information available to any one of them can affect the overall result.
What does effective oversight look like?
The changing nature of authorised payment fraud creates some important questions for governance and risk management.
Financial institutions already monitor fraud losses, suspicious transactions, customer claims and reimbursement performance. These are valuable measures, but they don’t necessarily tell management how much fraudulent activity is being prevented or where opportunities to intervene are being missed.
A reduction in fraud losses could reflect better detection, changing customer behaviour or a temporary shift in criminal activity. Higher intervention volumes might suggest improved monitoring, but could also indicate poorly calibrated controls creating unnecessary friction for legitimate customers.
Management needs to understand what sits behind those numbers.
The quality of fraud reporting therefore matters. It should help identify which types of scams are causing losses, where controls are proving effective, how often interventions prevent fraudulent payments and where suspicious activity is being recognised too late.
Receiving-account activity is another area worth examining, particularly where the same weaknesses or patterns appear repeatedly.
There is also a role for independent assurance in testing whether fraud prevention arrangements operate as expected, including how decisions are made, when concerns are escalated and whether lessons from confirmed cases lead to meaningful changes.
For boards and risk committees, the issue is how much confidence they can place in the organisation’s ability to recognise and respond to changing fraud techniques.
That confidence needs to take account of the information firms can access, the limitations of their controls and the extent to which they rely on intelligence held elsewhere.
The September payment diversion figures and October investment fraud warning illustrate how criminals can influence payment decisions without necessarily compromising the systems used to process them.
As those methods develop, financial institutions will need to keep examining whether their fraud prevention arrangements are looking for the right warning signs, and whether those signs are being recognised early enough to make a difference.
A payment can be correctly authorised and still result in a substantial fraud loss.
The challenge for financial institutions is understanding how much of the deception leading to that payment they can realistically identify, and whether their controls are making the best use of the information available.
Stay up to date with the latest news and views from the world of governance, risk, audit and compliance >>>





