Why Financial Regulators Are Now Stress-Testing AI Systems

The Bank of England has begun testing how artificial intelligence could impact the stability of the financial system, marking a shift in how regulators view AI inside critical infrastructure.

Rather than focusing only on how AI can improve efficiency in banking operations, regulators are now assessing how AI systems themselves might introduce new forms of systemic risk when used across financial markets and institutions.

This includes concerns about how widely deployed AI models could influence decision-making processes, how correlated outputs across institutions might amplify market movements, and how dependency on shared models or infrastructure could create points of systemic vulnerability.

Why this shift matters

Financial systems have always been designed around risk management, but most of that framework was built around human decision-making and deterministic software systems.

AI introduces a different type of behavior.

It is probabilistic, adaptive, and often opaque in how it produces outputs.

As AI becomes more embedded in trading, lending, fraud detection, and operational decision systems, it begins to operate across multiple institutions at once rather than within isolated environments.

This creates a situation where the same underlying models or similar model behaviors can influence decisions across the entire financial ecosystem.

From operational tool to systemic factor

Traditionally, technology in finance has been treated as an operational layer, supporting processes like payments, reporting, and compliance.

AI changes this relationship because it can influence decisions directly rather than just executing predefined logic.

That shift means regulators are no longer only concerned with system performance or security, but also with how AI-driven decisions might propagate across institutions.

This is why central banks are starting to treat AI as a potential systemic factor, similar to liquidity risk or market concentration.

Why this is difficult to regulate

One of the challenges is that AI systems are not centralized.

They are:

• deployed across multiple vendors
• embedded in different financial workflows
• updated frequently and independently
• influenced by shared datasets and architectures

This makes it difficult to fully map where AI is being used and how its outputs might interact across the system.

As a result, risk does not come from a single AI system failing, but from many systems behaving similarly at scale.

What this signals for critical infrastructure

The financial sector is often the first to formalize emerging risks, and this development reflects a broader trend.

As AI becomes more embedded in infrastructure systems, it starts to behave less like a tool and more like a shared dependency across institutions.

This raises questions that extend beyond finance:

• how should AI systems be monitored at scale
• how do you assess risk when outputs are probabilistic
• how do you prevent correlated behavior across organizations

The key shift is not that AI is being used in finance.

It is that AI is now being treated as something that can affect the stability of financial systems themselves.

That changes how it is regulated, monitored, and ultimately integrated into infrastructure.

At Lenet, we help organizations understand how emerging technologies like AI affect the structure and visibility of modern IT systems.

As AI becomes embedded across financial and operational infrastructure, the challenge is no longer adoption but understanding how systems interact at scale.

Our focus is on helping organizations maintain clarity and visibility as complexity increases across connected systems.