Failures involving external technology providers have caused more disruption for euro zone banks than fraud attempts, according to the European Central Bank.

Malicious activity made up a relatively small proportion of major cyber incidents at euro zone banks from the beginning of 2025 to September 2026, and most such cases had limited effects, the ECB said.

Access deeper industry intelligence

Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.

Find out more

Roughly one-fifth of reported cyber incidents stemmed from third-party suppliers, particularly providers of banks’ IT systems, the central bank said in a chapter of its Financial Stability Review.

It further noted that relying on large external providers could strengthen banks’ resilience because these “often have privileged access to highly cyber-capable frontier models and possess the expertise to make them work”.

However, it warned that this reliance could also create problematic “dependencies”, as major providers are often headquartered outside the European Union.

The ECB called for coordinated action by banks, technology companies, and public authorities to improve defences against AI-powered attacks.

Its recommendations included advising customers to retain “sufficient cash” in the event that their bank is disrupted.

Rapid improvements in the cyber capabilities of AI models are increasing cyber risks, the ECB said.

These include the use of agentic attacks, a broader attack surface, and the possibility that AI agents may behave in ways not intended by their operators.

Although defenders can also use AI tools, they may face a structural disadvantage in the short term.

The ECB identified two principal risks: first, that advances in AI-enabled cyber capabilities could exceed organisations’ ability to respond, potentially overwhelming defences; and second, that continued development of AI models could give a wider range of threat actors more sophisticated capabilities, raising the likelihood of more serious incidents.

The central bank added that geopolitical disparities and competition over the development, access, and governance of advanced AI models could intensify these risks.

Restrictions on, or the withdrawal of, access to the most capable AI models or related critical ICT services for geoeconomic reasons could create substantial operational challenges.