Lloyds expands agentic AI platform for debit card fraud investigations after 10,600-case trial

Lloyds Banking Group said in a press release that it is expanding its agentic AI platform for fraud support after the tool supported fraud specialists across more than 10,600 live customer cases in 90 days.1

The platform helps colleagues investigate suspected debit card fraud, according to the press release, and uses a group of specialist AI agents that each support a different part of an investigation.1 A central system coordinates the agents and guides fraud specialists through up to nine stages of a case, Lloyds said.1

Lloyds said overall decision accuracy passed 90% in September, and that in cases where the platform identified suspected fraud, more than 90% were later confirmed as fraud.1

Lloyds tested outputs against completed case outcomes and used specialist feedback, delivering 50 improvements through seven updates over 90 days; it said this led to fewer false alarms and missed fraud cases but published no figure for missed fraud.1

The platform has roughly halved the number of systems specialists use on a call, down from 12 to 15, and Lloyds aims to cut this to two or three by the end of the year.1

Lloyds said fewer overrides mean colleagues increasingly use recommended decisions unchanged, while trained specialists review the evidence and stay accountable for the customer decision; it did not give an override rate.1