Germany’s financial regulator is stepping up its scrutiny of banks’ technology resilience, announcing a new programme of targeted IT inspections as concerns grow that advances in artificial intelligence could expose weaknesses in both modern and legacy banking systems.
The Federal Financial Supervisory Authority (BaFin) said it is creating a dedicated division to carry out ‘IT spotlight’ inspections, allowing supervisors to examine, test and monitor specific technology risks more quickly than through full-scale supervisory reviews.
The move comes as regulators increasingly focus on how financial institutions manage software resilience, cyber security and operational risk in an AI-driven environment.
For software testing and quality engineering teams, the announcement signals that regulators are becoming more interested not simply in whether banks have testing programmes, but whether those programmes can identify vulnerabilities quickly enough as AI changes the threat landscape.
BaFin president Mark Branson warned that cyber risks are “growing” and “substantial” because of advances in artificial intelligence.
He said AI models are capable of identifying and exploiting weaknesses in both new and legacy IT systems, adding that the new inspection approach would enable BaFin to respond more rapidly to emerging threats.
DORA guidance
The new inspections also reinforce guidance BaFin published earlier this year that places artificial intelligence firmly within operational resilience and ICT risk management rather than treating it as a standalone innovation issue.
The regulator said the guidance was intended to help firms apply the EU’s Digital Operational Resilience Act (DORA) when deploying AI systems across their technology estates. It warned that AI was already being used throughout financial services but that the implementation and operation of these systems could create significant ICT-related risks.
In BaFin’s view, AI systems belong within the same ICT risk management framework as any other critical technology asset.
The authority said the security and resilience of AI systems must be maintained throughout their entire lifecycle, from data acquisition and model development through to ongoing operation and eventual retirement.
That significantly expands the role of testing beyond release validation. Rather than viewing AI testing as a one-off approval before deployment, BaFin’s guidance positions continuous assurance as part of wider operational resilience.
The regulator stressed that established software engineering disciplines, including unit testing, integration testing and source code reviews, remain essential for validating AI systems. It also said testing should be proportionate to the criticality of the business functions being supported.
BaFin highlighted that generative AI presents additional testing challenges, including the possibility of unannounced model changes when organisations depend on third-party AI services.
The guidance also extends software assurance into cyber resilience. BaFin pointed to adversarial testing techniques, including simulations of data poisoning and evasion attacks, alongside penetration testing designed to identify AI-specific vulnerabilities.
The regulator warned that backdoors could be introduced into models during training, while insecure deployment practices could allow attackers to steal or manipulate systems, potentially leading financial institutions to make incorrect decisions.
The emphasis on lifecycle testing is echoed by industry observers.
Marina Marusenko, a risk manager at ING, wrote on LinkedIn that “BaFin treated AI as an ICT risk issue, not an innovation topic,” adding that this distinction mattered because “the same identification, protection, detection, and recovery requirements applied, with no separate track and no special exemption.”
She also said the guidance made board accountability explicit, arguing it went beyond approving strategy documents and required ongoing engagement with technically complex systems.
That means responsibility for AI quality can no longer remain within data science teams but must become part of enterprise-wide operational resilience.
Marusenko said lifecycle management ran throughout the guidance, covering development, deployment, monitoring, change management and decommissioning.
“A one-time approval at go-live was not sufficient,” she wrote, arguing that AI systems required continuous oversight around model drift, data quality and version control.
The guidance also places significant emphasis on ICT third-party risk. BaFin noted that many AI systems depend on cloud providers and a relatively small number of external vendors, highlighting the importance of assessing concentration risk, portability, audit rights and credible exit strategies.
Marusenko said institutions using large language models through external APIs should pay particular attention because concentration risk and subcontracting chains could undermine resilience if not properly governed.
She also pointed to the growing risk of “shadow AI,” noting that applications integrating external AI models through APIs could transform software originally intended as non-AI into AI systems without organisations having full visibility.
Although BaFin described the AI guidance as non-binding, both the regulator and industry observers presented it as a blueprint for how DORA should be applied to AI systems.
Combined with BaFin’s new programme of targeted IT inspections, the direction of travel is becoming increasingly clear. Regulators are moving beyond high-level governance expectations and towards more focused examinations of how banks test, validate and monitor increasingly AI-enabled technology environments.
For banks, that raises the importance of maintaining robust regression testing, vulnerability testing, adversarial testing, change management and evidence that critical systems remain resilient following software releases, model updates and infrastructure changes.
16 SEPTEMBER IN LONDON
REGISTER TODAY – SIMPLY CLICK HERE
Why not become a QA Financial subscriber?
It’s entirely FREE
* Receive our weekly newsletter every Wednesday * Get priority invitations to our Forum events *
REGULATION & COMPLIANCE
Looking for more news on regulations and compliance requirements driving developments in software quality engineering at financial firms? Visit our dedicated Regulation & Compliance page here.
READ MORE
- Inside Rabobank: Engineering resilience by design
- Can AI agents finally automate data testing?
- Continuous testing drives DORA compliance
- Why software testing may face a major rethink
- Buy or build? AI rewrites software testing for banks
WATCH NOW
QA FINANCIAL PODCASTS
CLICK HERE TO LISTEN TO OUR EXCLUSIVE CONVERSATIONS
