The Bank of England is paying close attention to potential risks of AI on markets in a recent policy meeting . The central bank, which is responsible for keeping economic markets stable, is now asking what happens when AI shares drop causing leveraged investors to run for the door, in turn causing debt backed by data center dreams to sour, and scenarios in which frontier models make cyber attacks faster than banks can patch their systems. This is pointing to AI playing a much bigger role in economic planning than other technology waves have in the past.

While a fund manager can buy the dip if markets collapse and then sell the rally, a central bank has a different job. It looks for the point where big market moves can have serious negative and destabilizing impacts on the economy. The Bank of England now sees that point approaching with AI.

Its Financial Policy Committee said AI related shares have helped push global equity markets higher, with gains driven by a narrow group of companies. It flagged rising index concentration, heavier hedge fund leverage, retail flows into exchange traded funds and rapid growth in leveraged ETFs. If earnings expectations crack, the Bank warned, those same forces could turn a short-term tech selloff into a much bigger and longer-lasting market event.

The New Risk Is Not Just Valuation

Investors and market-makers argue over AI multiples and whether company valuations are far outpacing their ability to generate revenues or otherwise realize value in the long run. This leads central bankers to ask who is financing the build out, and the long-term stability in doing so.

The Bank of England said AI companies’ use of debt and credit markets has accelerated with greater use of public debt, private credit, leveraged finance and structured finance. It called the pace of investment “unprecedented historically.” While the amount of debt is still manageable currently and there’s no sign of immediate danger, the banks are highlighting the dangers of the financing of the AI industry. AI is not only a software story, but also increasingly about construction, power, manufacturing and investors’ need for returns.

Equity investors can absorb large losses without pulling the banking system into trouble, but credit adds lenders, covenants, collateral values, refinancing dates and off balance sheet structures. A data center can look like a gold mine in a spreadsheet until power costs rise, demand fluctuates, technology ages faster than expected or a borrower needs fresh cash in a tighter market.

The Bank says AI may raise productivity and lift long term growth. But the worry is timing. Markets have basically paid in advance for a future filled with cheap compute, steady power, fast adoption and thick margins. Central banks now have to model the other case. The AI trade has a feature central banks dislike: it touches many different markets at once.

AI capital spending supports growth in regions tied to chips, cloud infrastructure and data centers. It shapes expectations for productivity. It affects equity indices, especially once highly valued companies are added to indexes and ETFs. It is now starting to show up in credit. The Bank of England said shocks to AI related assets could travel more widely through the financial system as the AI ecosystem becomes more dependent on external finance. A negative reset in earnings expectations could weaken growth forecasts and even ripple into sovereign debt markets, especially where debt paths already look strained.

Cyber Risk Gives AI Another Potential Danger

The Bank of England’s warning is not limited to asset prices. The FPC said recent gains in frontier AI capabilities have increased risks tied to cyber and operational resilience. AI systems may find and exploit software flaws faster than banks, exchanges, vendors and cloud partners can fix them. This means firms face more patching, more testing and more strain on technology teams. Without due care, vulnerabilities pile up and a systemic cyber event becomes more likely.

Europe is treating that risk with unusual urgency. The European Systemic Risk Board warned on July 7 that frontier AI models could raise the speed, scale and sophistication of cyber attacks, at least in the short to medium term. It also said concentration of leading AI providers outside the European Union creates strategic dependency and geopolitical risk.

The European Central Bank went further. In a July 7 letter to significant euro area banks , ECB Banking Supervision said emerging AI models can identify vulnerabilities and generate working exploits at “unprecedented speed.” It told banks to submit action plans by Oct. 31, 2026, covering faster patching, better monitoring, AI enabled defensive tools, third party risk checks, cyber hygiene, older technology and recovery planning.

Regulators Are Building The Rulebook In Public

The Financial Stability Board is trying to create common ground so that AI progress can be achieved while at the same time maintaining market and cyber stability. In June, it proposed 12 sound practices for financial institutions adopting AI, aimed at firm wide governance, senior management oversight, risk controls, model life cycle checks, cyber risk and third party exposure. It described AI adoption as a source of opportunity, but warned that rapid use can amplify or introduce risks that firms must identify and manage.

In May, the IMF said AI is changing how the financial system deals with vulnerabilities and incidents, and that extreme cyber losses could trigger funding stress, solvency concerns and wider market disruption. It pointed to the system’s shared digital backbone of software, cloud services, payments networks and data links. A flaw in one widely used component can become a sector problem fast.

The Bank of England had already laid out its own AI monitoring approach in April. It named four areas to watch covering AI in core bank and insurer decisions, AI in trading and investment, reliance on outside AI service providers, and the changing cyber threat. It also said heavy use of similar AI tools in markets could lead firms to take correlated positions during stress. That is central banker language for a familiar market accident in which too many models that say sell at the same time, causing a “flash drop” in the market.

While banks are sounding warning bells when it comes to market and technology vulnerabilities with AI, None of this kills the AI investment case. AI may still cut costs, improve drug discovery, speed software work, help banks detect fraud, improve customer service and lift productivity.

As AI matures and finds deeper use in organizations, AI is becoming a financial stability subject before it has finished becoming a profit story. Many companies still have to prove that customers will pay enough for AI services to justify the huge buildout behind them. Many lenders still have to prove they understand the collateral. Many banks still have to prove they can survive a world where attackers use machines to find weak spots at machine speed. So while some in the market are calling for a pending AI bubble, bankers are taking active steps to ensure that if the bubble does pop, economies are well prepared.