AI-Enabled Fraud Grows More Complex Across EMEA Banks

Across Europe, the Middle East and Africa, banks say fraud is becoming markedly harder to detect, and AI now sits on both sides of the fight.

AI-enabled fraud in EMEA is becoming harder to detect faster than it is growing in size, and the institutions keeping pace are those linking their fraud defences across the whole enterprise. More than a third of banks, neobanks and fintechs across Europe, the Middle East and Africa have seen fraud attempts climb by more than 25% over the past 24 months, according to new research from FICO, the analytics company behind the FICO Score, which it says is used by 90% of top US lenders.

The study, carried out with the research firm Finextra and published as Fraud in the Age of AI: Insights from EMEA, surveyed 202 senior fraud, risk and technology professionals at retail banks, neobanks and fintechs. The regional findings stand out: 52% of EMEA institutions now cite rising fraud complexity as their single biggest prevention challenge, against 40% of firms globally.

How bad is the fraud problem becoming in EMEA?

The headline numbers point to pressure building on two fronts. Across EMEA, 36% of institutions report that fraud attempts have risen by more than 25% in the past two years, while 28% say the financial losses from fraud grew by more than 25% over the same period. What stands out, FICO says, is the difficulty of the attacks themselves: 52% rank complexity as their top challenge, a full 12 percentage points above the global average.

Why does complexity worry EMEA banks more than scale?

EMEA fraud leaders consistently rate emerging threats as more dangerous than their global peers do. On FICO's five-point scale, respondents in the region scored AI-enhanced fraud at 4.42, against a global average of 3.86, and social engineering scams at 4.36, against 3.90. Synthetic identities, at 4.02, and deepfake fraud, at 3.96, also ranked above global concern levels.

External data helps explain the unease. According to Signicat, the European digital identity firm, deepfake fraud attempts across the region rose by 2,137% in the three years to early 2025, moving from roughly 0.1% of detected attempts to about one in 15. Signicat's The Battle Against AI-Driven Identity Fraud report also found that 42.5% of fraud attempts in the financial sector now involve AI, even though only 22% of institutions had deployed AI-based prevention tools at the time of its research.

Can EMEA banks deploy the AI they are investing in?

The same tools that worry fraud teams are the ones they are racing to adopt, and that is where many are stalling. Nearly half of EMEA respondents (48%) rank effective AI integration as a top challenge, almost identical to the global figure, which suggests that embedding AI into existing fraud controls is as much an operational hurdle as a technical one. The strain shows up in customer experience too: 51% of EMEA institutions report false positive rates above 15%, meaning legitimate customers are being flagged and delayed even as controls tighten.

Who is really making EMEA fraud decisions now?

Increasingly, it is outside vendors. Third-party vendor models now carry the greatest influence over fraud decisions at 46% of EMEA institutions, nearly double the 24% that lean mainly on in-house models, a sign of how far banks are reaching for external expertise to keep up. At the same time, 86% of institutions see a unified, enterprise-wide fraud strategy as essential, though only 38% treat it as a critical near-term priority, with a further 48% pursuing it as a longer-term goal.

What should fraud teams prioritise next?

Adam Davies, FICO's vice president of product management, sets out four priorities. The first is moving AI from pilot projects into full production within disciplined control frameworks. The second is investing in active orchestration so that signals are shared across models, channels, products and portfolios in real time. The third is applying risk-based prioritisation led by financial impact and regulatory exposure. The fourth is treating fraud prevention and customer experience as one connected problem rather than two competing demands.

Davies notes that confidence in agentic AI is particularly high in EMEA, and argues that the next 24 months will be shaped by how well institutions connect in-house models, vendor intelligence and consortium data into a single real-time view. FICO's own detection models draw on tagged transactions from a consortium of more than 10,000 financial institutions, the kind of shared data pool the research suggests will matter more as fraud grows more sophisticated.

Why This Matters to FinanceX Readers

For banks and investors, the signal in this research is about operating model as much as threat level. EMEA institutions have built credible AI detection foundations, yet the survey suggests the next competitive gap will open between firms that can orchestrate their data and those that keep buying standalone tools.

For fraud and risk leaders, the practical takeaways are to close the gap between AI pilots and production, to measure success by financial and regulatory outcomes, and to treat customer friction as a genuine cost of fraud control. For investors watching the regtech and fraud-prevention space, the rising reliance on third-party vendor models, now the dominant influence on EMEA fraud decisions, points to continued demand for consortium data and orchestration platforms over single-point products.