More than half a billion pounds were lost to scammers last year as victims are being manipulated by artificial intelligence.

AI has lowered the 'barrier to entry' for criminals, with fraud losses involving the technology surging by nearly a fifth.

It has helped criminals to make communications appear legitimate, send out messages at scale and mimic people's loved ones, banking body UK Finance said.

In four out of ten cases, authorised push payment (APP) scam victims lose their money for good.

The jump in APP fraud losses reflects a growth of complex scams that manipulate people into transferring money for counterfeit goods, services

or opportunities, the organisation said.

APP fraud losses rose sharply to £576.4million, the highest since 2021 during the pandemic, when £583.2million was stolen.

UK Finance said 248,070 cases were recorded, a 7 per cent annual increase.

Ruth Ray, managing director of economic crime at the firm, said the body is shocked, but not surprised by the figures.

She said: 'We are well aware that, particularly with the advent of AI, it lowers the barrier to entry for criminals to make their scams more sophisticated and make more of us vulnerable to and susceptible to clicking on links and content.' 

Purchase scams accounted for 71 per cent of cases, with losses at £118.1million.

Investment fraud made up the highest share of losses, at £221.5million. Romance fraud losses increased by 23 per cent to £39.2million. But impersonation fraud losses fell by 12 per cent.

This occurs when criminals pose as organisations such as a bank or the police.

In 2025, banks paid £354.3million to victims of APP fraud, equivalent to 61 per cent of losses.

In October 2024, APP fraud reimbursement rules came into force, meaning banks must repay victims, except in cases of customer gross negligence.

This applies when a transfer is made to and from a UK bank account and there is an £85,000 reimbursement limit, although banks may choose to repay more.

Last year, banks blocked £1.68 billion in unauthorised payment fraud, equivalent to 70p in every £1 of attempted attacks.

Despite this, authorised losses remain high as criminals use new technologies to exploit victims.

Jonathan Frost, global advisory director at BioCatch, said: 'Financial institutions must detect social engineering in real-time, intervening before funds leave an account, and disrupt the mule accounts that fraudsters rely on.'