- 92% of European banks have adopted AI, but scaling remains a major challenge.
- Unsupervised AI is revolutionizing fraud detection, cutting investigation time by up to 70%.
- BBVA centralizes 4,000 tech staff in AI Transformation and reports significant productivity gains.
- Regulatory pressure and customer trust are key factors for sustainable AI adoption.
The era of artificial intelligence as a mere buzzword in banking is over. According to the European Banking Authority, 92% of European banks have already integrated AI into their operations, while the remaining 8% are either testing or discussing use cases. This isn’t just about chatbots or simple automation; it’s about embedding AI into the very fabric of financial services, from risk management to customer experience.
But there’s a catch. While adoption is widespread, scaling AI from isolated pilots to enterprise-wide processes remains a massive hurdle. A global study shows that 88% of organizations use AI in at least one function, yet only a third have started scaling their programs. In the banking sector, this gap is even more pronounced due to regulatory scrutiny, legacy systems, and the need for explainability. This article dives into how banks are tackling these challenges, with a special focus on fraud detection and BBVA’s ambitious AI transformation.
From Pilots to Core Processes
The transition from pilot projects to critical business processes is where the real value lies. In banking, the most advanced use cases are in fraud prevention, risk analysis, credit scoring, regulatory compliance, document automation, and back-office efficiency. AI can spot anomalous patterns, speed up processes, and improve response times. However, scaling requires explainable models, bias controls, traceability, integration with legacy systems, and a clear return on investment. In insurance, the picture is similar: EIOPA reports that 50% of non-life insurers and 24% of life insurers already use AI in pricing, underwriting, fraud, or claims management. The potential is huge, but data quality, fragmented sources, and customer trust remain obstacles.
Banks are also realizing that AI isn’t just a tech project; it’s a strategic capability. The European AI Act and DORA are pushing institutions to treat AI as a governed, audited, and monitored asset. This means building robust governance frameworks and ensuring that AI decisions can be explained to regulators and customers alike.
Regulation, Governance, and Customer Trust
Regulatory pressure is mounting. The AI Act introduces a risk-based approach, while DORA strengthens digital operational resilience. In practice, banks must now demonstrate that their AI systems are safe, transparent, and accountable. But it’s not just about compliance. Customer trust is equally critical. According to the Qaracter Barometer, 81% of users believe AI can improve financial management, yet 54% are concerned about its use. People see the value but demand transparency, security, and human oversight when decisions are significant. Banks that strike this balance will win the race.
Marta Sanz, Global Director of Exponential Technologies at Qaracter, puts it bluntly: ‘AI can no longer be managed as a separate innovation lab. The real leap happens when models are integrated into critical processes with data governance, risk controls, cybersecurity, impact metrics, and human supervision.’ This sentiment echoes across the industry.
AI in Fraud Detection: A Game Changer
One area where AI is making a tangible difference is internal fraud detection. The Association of Certified Fraud Examiners (ACFE) reports that internal fraud costs Latin American banks up to $200,000 per case, and traditional systems can take around two months to detect it. In a study of 108 cases across the region, 37 were in Mexico, 18 in Brazil, and 11 in Colombia. Corruption was the most common scheme, appearing in 61% of cases. Interestingly, 54% of these frauds were caught thanks to employee tips, while only 2% were detected through data monitoring systems. That’s a huge gap.
Unsupervised AI changes the game. Instead of relying on predefined rules, it learns each user’s normal behavior and flags deviations in real time, even for never-before-seen fraud patterns. Eduardo Álvarez, an expert at Nubatech, explains: ‘Traditional systems are rule-based: if X happens, alert. But rules are only as good as the fraud we already know. AI can spot anomalies that humans would miss.’ This technology can reduce investigation time by up to 70% and provides explainable alerts, which is crucial for regulatory compliance.
BBVA’s AI Transformation Journey
BBVA is a prime example of a bank going all-in on AI. The Spanish lender has created a global AI Transformation unit, bringing together 4,000 employees with tech and data profiles under the leadership of Antonio Bravo, who reports directly to the president. This unit aims to industrialize AI and accelerate the deployment of AI agents. BBVA has also rolled out AI tools to over 100,000 employees, with 70% using them monthly. Employees report saving three hours per week on routine tasks, according to internal estimates.
The bank’s strategy is built on three ‘waves’: adoption, a set of six specialized robots and two tech pillars called ‘The Eight’, and a framework called ‘The Framework’ to scale and systematize AI development. In its Corporate and Investment Banking (BEI) division, more than 7,000 professionals already use AI tools. For instance, Blue Enterprise is an intelligent assistant available 24/7, while AI Banker helps relationship managers prepare for client meetings, saving an hour per visit. AutoPF, a tool for financial program preparation, has been used by 86% of BEI managers in Spain, with over 900 active users and 34,000 conversations started.
BBVA is also investing heavily in technology, with around €4 billion earmarked for 2025. The bank is working on integrating commercial and risk analysis capabilities to speed up decision-making and reduce administrative burdens. Jaime Sáenz de Tejada, global head of BEI, says: ‘We are moving from individual AI tools to the full transformation of our commercial, risk, and customer service processes. The opportunity is not just to be more efficient, but to better accompany companies and anticipate their needs.’
All these developments point to a clear trend: AI is no longer a nice-to-have in banking; it’s a must-have. From the 92% adoption rate in Europe to the specific fraud detection breakthroughs and BBVA’s massive organizational overhaul, the industry is moving from experimentation to industrialization. The banks that will thrive are those that can scale AI responsibly, with strong governance, clear metrics, and a human touch. As the saying goes, the future is already here—it’s just not evenly distributed. But in banking, it’s spreading fast.