BioCatch, a fraud prevention and financial crime detection firm that uses behavioural biometric intelligence, has published new research revealing that UK banks are increasingly exposed to AI-driven fraud that is becoming progressively harder to distinguish from legitimate customer activity.
The UK findings draw on survey responses from 80 fraud management, financial crime prevention, and risk and compliance professionals. Among the headline results, 84% of respondents believe AI has raised the sophistication of fraud and scam schemes, while 75% say differentiating between legitimate AI-assisted banking and malicious or manipulated activity will pose a very significant challenge. Separately, 73% said fraud attempts at their institution are on the rise and 64% reported growing fraud losses. A further 76% indicated that real-time intelligence on receiving accounts would meaningfully strengthen their capacity to identify and prevent scams.
These results emerge as UK financial institutions face escalating pressure from scams, mule account networks and increasingly refined social engineering tactics, compounded by the operational burden of mandatory reimbursement obligations for authorised push payment fraud. The rise of agentic AI intensifies this challenge by requiring banks to determine whether a given transaction reflects genuine customer intent, legitimate automated activity, or criminal manipulation before any funds are transferred.
The UK data forms part of a broader global survey covering 1,440 fraud management, anti-money laundering (AML), and risk and compliance leaders across banks in 25 countries on five continents. Globally, 84% of respondents identify AI agents as the most exploitable vulnerability facing the industry in the coming year. An additional 88% say AI has already elevated the sophistication of fraud, 60% expect AI-mediated banking to erode the effectiveness of conventional fraud defences, and 72% believe distinguishing legitimate AI-assisted transactions from malicious ones will become very difficult in a world where AI agents routinely initiate banking activity.
The global data also points to a marked deterioration in fraud conditions over the past year. In 2025, 71% of respondents reported rising fraud attempts at their organisations; in 2026, that figure climbed to 81%. The proportion reporting year-on-year increases in fraud losses rose from 59% to 76% over the same period. Close to half of respondents this year said their organisation loses more than $10m annually to fraud, with 20% reporting losses exceeding $25m and 5% suffering more than $50m in annual losses.
Survey participants, all manager-level or above with 79% at director level or above and 23% in their bank’s C-suite, also indicated a strong desire for greater cross-institutional collaboration. Globally, 86% said gaining real-time intelligence on receiving accounts in interbank transactions would improve their ability to prevent scams, and 85% said interbank intelligence sharing would help combat fraud and financial crime more broadly. The importance of receiving-account intelligence is underscored by the fact that many scams originate outside the banking session itself, while stolen funds are typically moved rapidly through mule accounts.
In the UK specifically, 61% of respondents expect the wider adoption of AI-mediated banking to reduce the effectiveness of existing fraud signals, marginally above the global average of 60%. Meanwhile, 80% of global respondents said their institution has already encountered attacks making use of agentic AI, and 76% expressed strong concern about the accelerating pace of fraudulent activity. Customer trust also emerged as a significant factor, with more than 96% of respondents saying their institution already monitors customer attrition linked to fraud and scam experiences, and 39% identifying it as a primary driver of investment decisions. Globally, 68% of banking leaders believe their organisation’s approach to fraud prevention and reimbursement has resulted in a net loss of customers, with 56% attributing this to unreimbursed losses and 44% to excessive friction applied to genuine customers.
UK respondents reported comparatively lower rates of claim rejections and customer losses than their global counterparts. Only 19% of UK respondents said their organisation rejects more than $5m in fraud claims annually, against 59% globally, and just 14% said their bank’s customers lose more than $10m to fraud and scams each year, well below the global average of 39%.
BioCatch global advisory, EMEA, Jonathan Frost said, “Agentic AI is making fraud faster, more scalable, and harder to detect. UK banks should prioritise early prevention in the payment process. Although strong reimbursement rules protect victims after fraud, they do not prevent emotional harm, disrupt mule accounts, or stop criminals from profiting. Criminals will inevitably use AI, potentially leading to exponential growth in fraud. Only behavioural insights, shared in real time across the sector, can detect customer manipulation, assess user intent, identify mules, and support a risk-based approach to friction.”
BioCatch CEO Gadi Mazor said, “AI is starting to reshape how customers interact with e-commerce sites and financial institutions and will change how criminals execute fraud and other financial crimes. As digital interactions continue to grow faster, more automated, and increasingly driven by agents, we must move beyond static identity checks and toward a deeper and immediate understanding of behaviour, intent, and trust.”
Copyright © 2026 FinTech Global
Copyright © 2018 RegTech Analyst
