Metro Bank customer loses £14,000 to fraud that used AI chatbot Claude to drain his account

Key points

• Zoli Rutter, a businessman from Sussex, had £14,244 taken from his Metro Bank account without his authorisation.
• Fraudsters used the stolen money to buy credits for Claude, an AI chatbot made by the company Anthropic.
• Rutter says he told Metro Bank the money was being taken without his permission, yet the withdrawals continued.
• UK payment law may oblige Metro Bank to refund him in full.

Zoli Rutter runs a business in Sussex. He's not a careless man with money. Yet fraudsters got into his Metro Bank account and quietly emptied it, spending £14,244 on credits for Claude, an AI chatbot (a computer program you can have a text conversation with) made by the San Francisco company Anthropic.

Rutter spotted it and warned the bank. The withdrawals kept coming.

What actually happened?

Fraudsters bought credits for Claude using Rutter's account without his knowledge or consent. Credits are prepaid tokens that let users send messages to the chatbot and receive replies. Someone was spending Rutter's money on them. The Guardian first reported the story: Rutter contacted Metro Bank once he noticed the unauthorised transactions, expected the payments to be frozen, and watched the total climb past fourteen thousand pounds instead. Metro Bank hasn't publicly confirmed the exact sequence of events.

Should other bank customers be worried?

This specific scam isn't widespread, but it points to a pattern worth knowing. AI services that sell credits or subscriptions are now common targets for fraudsters testing stolen card details, because small repeat purchases can slip past basic fraud filters. We first picked up the connection between chatbots and consumer harm in our 15 July story on a customer trapped in automated dead ends; fraud is a sharper edge of the same problem.

If you see a charge you don't recognise, report it to your bank immediately and ask them to block the merchant as well as reverse the payment. One phone call isn't always enough. Follow up in writing, by app message or email, so you have a timestamped record.

Banks in the UK are bound by rules from the Payment Systems Regulator that require them to reimburse victims of authorised push payment fraud, where a customer is tricked into sending money themselves. Unauthorised fraud, where money leaves without the customer doing anything, carries even stronger legal protections under the Payment Services Regulations 2017. Rutter's case appears to fall into that second category, which means Metro Bank may be legally required to refund him.

What happens next?

Rutter's pushing for a full refund. If Metro Bank refuses, he can escalate to the Financial Ombudsman Service, a free independent body that resolves disputes between customers and financial firms in the UK.

The bigger point is this: AI services are now woven into everyday financial fraud, not as the attacker but as the destination for stolen money. Claude didn't do anything wrong. The gap was in the bank's fraud detection, and it's a gap that cost one person £14,244.

Check your statements today. One unknown charge is worth five minutes of your time.

Common questions

Can the bank be forced to pay Rutter back?

Under UK payment law, banks must generally refund customers for transactions they didn't authorise, unless the bank can prove the customer was grossly negligent. Rutter reported the fraud while it was happening, which makes that argument very difficult for Metro Bank to sustain.

Is Claude or Anthropic responsible for the fraud?

No. The chatbot was simply where the stolen money was spent. Anthropic isn't accused of any wrongdoing.

What should I do if I see a charge I don't recognise?

Call your bank, report it as unauthorised, ask them to block the merchant and issue a new card. Then follow up in writing through the bank's app or email so you've got a timestamped record of your report.