European banks see more complex fraud; AI controls lag

A survey of 200 European fraud leaders, conducted by Finextra in association with NICE Actimize, finds banks facing rising fraud complexity across payment channels and difficulty integrating AI controls quickly enough to match new threats.

The survey was presented in a Finextra webinar moderated by researcher Sharon Kimathi. Respondents reported fraud patterns in 2026 expanding beyond card fraud and account takeover to include payment flows initiated with synthetic or AI-generated instructions and scams targeting virtual assets.

Many institutions are adopting artificial intelligence to detect and block fraud across channels. The survey found respondents view AI as necessary to broaden detection coverage and to identify novel attack patterns that rules-based systems miss. At the same time, legacy infrastructure, data quality gaps and the need to tune models for local markets are slowing AI deployment.

The responses showed uneven levels of fraud expertise, funding and tooling across European markets. Some banks reported stronger internal analytics teams and higher investment in real-time monitoring, while others rely more on third-party vendors or under-resourced teams. These gaps produce inconsistent controls and different capacities to respond to cross-border fraud schemes.

Fraud leaders reported pressure to prioritise limited resources as fraud accelerates across channels. Institutions face choices between strengthening core controls for card and account fraud, expanding payments screening and building specialised detection for virtual asset flows. Many respondents flagged model governance, data integration and faster decisioning as immediate needs to support advanced analytic tools.

Integration challenges were cited as a major barrier to scaling AI. Respondents noted that deploying AI requires changes to data pipelines, operational processes and compliance checks. Where banks have fragmented data or siloed teams, model performance suffers and pilots often fail to scale for lack of engineering resources or alignment among fraud, IT and compliance functions.

Respondents described evolving fraud tactics that include automated social engineering, realistic synthetic identities and payment structuring through non-traditional rails. Virtual asset fraud drew attention because of limited regulatory standards and the technical difficulty of tracking token flows.

Panel participants recommended investments in shared data sources, stronger cross-border cooperation and clearer model validation processes to accelerate safe AI deployment. The webinar aimed to help fraud, risk and compliance leaders set priorities for 2026 and beyond.

The survey maps the competing demands facing European banks: defend established channels, expand detection to newer payment types and change systems so AI-driven tools can be deployed at scale.