Lloyds Deploys Agentic AI to Tackle Fraud in Real Time

Lloyds Banking Group has deployed an agentic AI system that runs multiple AI agents at once to support its fraud teams in real time. It is the latest sign that autonomous AI is moving from pilot to production in UK banking.

“By combining agentic AI with human oversight, we can create an invisible forcefield around our customers, monitoring risk in real time and stepping in only when it matters,” notes Tom Martin, Business Platform Lead for Economic Crime Prevention at Lloyds Banking Group.

The system marks a shift in how the technology is applied. AI has assisted fraud prevention for years, but running several agents in concert, each handling a distinct task, is a step change for the UK’s biggest digital bank.

Many agents, one accountable human

During customer calls, multiple AI agents now operate behind the scenes at once. They carry out identity checks, transaction analysis and scam risk assessment in real time, giving colleagues faster and better-informed support.

Crucially, the agents advise but do not decide. The colleague on the call stays accountable and can overrule any suggestion the system makes. The agents simply surface risk earlier, flagging only the cases that warrant a second look.

That distinction matters. A wrongly blocked payment means an angry customer and a complaint to answer. A missed scam means a refund the bank has to cover.

The fraud agent was built by fraud, technology, data and risk specialists using Envoy, the group’s secure AI platform for developing and testing agents under what it calls rigorous oversight.

A £1bn fraud problem, by the numbers

The investment is being framed in hard figures. Lloyds prevented more than £1bn (US$1.35bn) of fraud in 2025 and has put £100m (US$135m) into new fraud technology since 2023.

Those numbers give the deployment a clear commercial logic. Every pound of fraud stopped is a pound not reimbursed, and faster agent-assisted decisions free frontline staff to handle more cases.

The push sits inside a far larger programme. Lloyds has invested more than £4bn (US$5.4bn) in digital transformation and is expanding AI across the business.

It is also extending its agentic tooling beyond fraud. The group is rolling out Microsoft’s agentic AI suite across its operations from June 2026, a sign of how central the technology has become to its plans.

Scam Check and a visible signal

A second tool aims the technology at customers directly. Coming shortly, Scam Check will be embedded into certain payment journeys across Lloyds, Halifax and Bank of Scotland.

Ron van Kemenade, Group Chief Operating Officer at Lloyds Banking Group, frames the work as part of a wider operating shift. “AI is now integrated across our organisation and is transforming how we operate and serve customers,” he says. “Our ongoing investment keeps Lloyds Banking Group ahead in responsible tech adoption within financial services.”

When a customer tries to pay someone new for an online purchase, Scam Check can ask a few simple questions. It then invites them to upload screenshots of the item, including from marketplace listings. Machine learning and image analysis then scan for recognised scam indicators and present a tailored warning if a risk appears.

Customers will also see the bank’s anti-fraud Dark Horse logo in the app and at the point of purchase, a visible signal that protection is running in the background.

The formula is simple enough to copy: let the agents do the digging, leave the decision to a human and let the customer see the shield go up