The opinions expressed here are those of the authors. They do not necessarily reflect the views or positions of UK Finance or its members.

The threat hasn’t eased; it’s just moved. Fraud is shifting shape, supported by off-the-shelf tools that make it faster to deploy and harder to spot. The conversation at recent industry events, including UK Finance’s Key Conversation: Fraud, reflected this. There’s a noticeable shift from reassurance to realism. Banks are adapting, but fraudsters are too, and often faster.

Most of the energy is still going into catching fraud once the money starts moving. But fraud doesn’t start there. That’s the finish line. By the time a transaction hits, the attacker’s already done the groundwork; recon, setup, social engineering, session compromise. And the reason so much of it works? Losses remain the most reliable trigger for many fraud responses, and that time gap is still where much of the damage happens.

While many banks are investing in behavioural and biometric signals under SCA and PSD2, fraud prevention often still focuses on transaction-level activity. That leaves a window where pre-transaction signals go unused, especially in session manipulation, device interference, or social engineering setups. 

There’s growing interest in flipping that. We heard a lot about the need to “shift left” - catching signals earlier, before, during the session, before a payment or account change is authorised. It’s where Cleafy operates. Real-time predictive detection, not post-mortem alerts. It’s a shift in mindset: from chasing transactions to tracking the full journey. How attackers behave, how sessions change, how signals build. 

It’s also clear that bridging the gap between cyber and fraud is finally getting the attention it deserves. For a long time, these teams worked in silos: different tools, different data, different priorities. But that’s starting to change. We’re seeing growing momentum around fusing intelligence in real-time, not just in theory but in practice. At BCC Iccrea Group, a major Italian bank, that shift led to an 80 per cent drop in fraud and over €0.5M saved each month, by spotting threats earlier in the session and giving both teams the signals they needed to act fast.

Another thing we couldn’t ignore: fraud teams are under pressure. The pace is relentless, but the resources haven’t caught up. Most teams are running rules that are outdated in fast-moving environments, and are expected to keep up with attackers who now operate like agile tech startups.

During a panel at the UK Finance event, one speaker noted: “Nobody grows up wanting to fight fraud; most of us fall into it.” That line landed. It spoke to how reactive fraud teams have had to be, and why the shift to earlier detection is finally gaining ground. 

Fraud is moving faster than most banks can respond. It’s being run like a business. Until it’s tackled with the same mindset, the gap won’t close; it’ll widen. But there’s some optimism too. The conversations are changing. The urgency is there. What matters now is action: spotting signals sooner, backing teams with the tools that keep up, and building systems that aren’t just reactive, but ready.