UK APP Fraud Losses Rise 19% as Payment Providers Face Growing AI Security Challenge

By Gemma Rolfe Fraud & Security
views

Financial losses from authorised push payment (APP) fraud in the UK increased by 19% in 2025, while unauthorised fraud losses declined by 5%, according to figures from UK Finance.

Envato

UK APP Fraud Losses Rise 19% Amid AI Challenges

The contrasting trends highlight a shift in criminal tactics, with fraudsters increasingly exploiting social engineering and artificial intelligence to persuade consumers to authorise fraudulent transactions rather than attempting to circumvent traditional payment security systems.

New research from Enfuce suggests consumers are increasingly willing to accept additional security measures, although significant concerns remain about the use of AI in fraud prevention.

Fraudsters Shift Towards Human Manipulation

Traditional payment fraud largely involved stolen credentials, counterfeit cards and unauthorised account access. Technologies including chip and PIN, tokenisation and stronger authentication have progressively strengthened defences against these threats.

However, criminals are increasingly exploiting psychological manipulation, using AI-generated communications and sophisticated impersonation techniques to convince legitimate customers to transfer money.

Daniel Alter, Chief Risk Officer at Enfuce, argues that payment providers must fundamentally reconsider their approach to fraud prevention.

“Twenty years ago, criminals were cloning cards. Chip and PIN helped turn the tide, but fraudsters famously don’t stand still. When one door closes, they find another.”

Alter believes the industry must become better at recognising when genuine customers are acting under fraudulent influence, even when they have willingly authorised a transaction.

Consumers Willing to Sacrifice Speed for Security

Enfuce’s report, Frictionless Payments. Invisible Trust, based on a survey of 2,000 UK consumers, reveals considerable support for stronger payment protection.

An overwhelming 95% would accept at least one additional security measure in exchange for safer payments. Some 47% would accept approval before payment, while 43% would tolerate additional identity verification.

“People don’t resent friction when they can see what it’s for,” Alter argues, suggesting that a short pause or additional verification can be justified when customers understand the protection it provides.

The findings challenge the industry’s longstanding assumption that friction inevitably undermines the customer experience.

AI Fraud Detection Faces a Trust Deficit

While artificial intelligence offers opportunities to identify fraudulent behaviour, consumer confidence remains limited.

Enfuce found that 90% expressed concerns about AI-powered fraud detection, with 51% worried about incorrect decisions and 39% citing data privacy.

Furthermore, 45% would feel uncomfortable knowing AI rather than humans was responsible for detecting payment fraud.

Perhaps most revealingly, only 19% selected AI fraud detection as an additional security measure they would accept, despite overwhelming support for stronger protection.

This exposes a significant challenge for payment providers investing in increasingly automated fraud prevention.

Combining AI with Human Judgement

Alter acknowledges that greater security cannot come at the expense of accountability.

“Providers need to use these new tools carefully,” he argues, emphasising the importance of proportionate intervention and human oversight.

For banks, fintechs and payment processors, the challenge is balancing intelligent automation with transparent decision-making and consumer confidence.

As fraudsters increasingly exploit human behaviour, the next generation of payment security may depend less on eliminating friction and more on introducing it intelligently.

As Alter concludes, the industry must focus on making it harder for criminals to persuade consumers to surrender their money, using AI alongside human judgement rather than replacing it.

Comments

Post comment

No comments found for this post