Artificial intelligence is no longer just a technology story. In a warning to G20 finance ministers, Andrew Bailey, governor of the Bank of England and chair of the Financial Stability Board (FSB), puts AI near the center of a wider discussion about financial stability.
Bailey’s message is direct: global markets have handled recent stress, including the shock from the Middle East conflict, but the system remains exposed. Energy prices are volatile, interest rates have risen, and risky assets still have high valuations. Against that backdrop, AI investment has become a potential channel for the next market shock.
AI valuations are becoming a financial stability issue
The Financial Stability Board coordinates financial regulation across G20 nations, and Bailey’s letter frames AI as more than a sector-specific concern. The issue is not simply whether investors are optimistic about artificial intelligence. It is how that optimism sits inside a market already shaped by tight conditions, expensive risky assets, and concentrated exposure.
Bailey identifies inflated AI valuations as one of several pressure points, alongside fragile government bond markets and opaque private credit. His concern grows sharper when those valuations are paired with leverage. In plain terms, more investors are using borrowed money to chase gains in markets that may already be priced for strong expectations.
That can make markets look stable while prices are rising. But it can also make any reversal more severe. Bailey writes, "As we have seen multiple times in the past, rising leverage is a feature of a maturing financial cycle. While it can reinforce rising markets, it can also intensify declines when sentiment turns, as recent weeks have demonstrated,"
Borrowed money can spread losses faster
The warning focuses on several ways leverage is entering the system. Bailey points to leveraged ETFs and trend-following strategies drawing in more retail investors. He also highlights hedge funds with positions in both equities and government bonds.
That combination matters because stress in one part of the market may not stay there. If investors need to unwind positions quickly, pressure can move from one asset class into another. A problem that begins around AI stocks, government bonds, or private credit could become broader if the same investors or strategies are exposed across multiple markets.
The risk is amplified by high valuations and heavy market concentration. Bailey also flags the growing web of cross-investments between AI companies and hyperscalers. If one major AI company struggles, the impact may not be limited to that company’s shareholders. It could affect other large technology firms and then reach the wider market.
This is why AI valuations are now being discussed as a systemic risk, not just a debate about whether certain companies are overvalued. The structure around the investment boom matters: who owns the assets, how they are financed, and how connected those owners are to other markets.
Frontier AI changes the cyber risk equation
Bailey’s second major concern is frontier AI. These models are gaining autonomous capabilities, and the letter warns that they could also make cyber threats more dangerous. Bailey writes that frontier AI could "materially alter the speed, scale and economics of cyber risk,"
The implication is that attacks could become cheaper, faster, and more frequent. For financial institutions, that is not only an operational problem. Cyber incidents can damage trust, and trust is central to the financial system.
The risk is made larger by the way banks depend on a small number of large technology providers. If one major provider is successfully attacked, institutions in several countries could be affected at the same time. Bailey also warns that uneven cyber defenses between countries may become a vulnerability of their own.
Frontier AI is not only a threat. Bailey also recognizes that it can help defend against cyberattacks. But his argument is that defensive uses require preparation, and that technical progress needs to be matched by governance and readiness.
Rules for advanced AI remain uneven
One of the clearest gaps in Bailey’s warning is the lack of rules in many countries for developing, releasing, or deploying advanced AI models. That creates risks beyond finance, but the financial system is especially exposed because of its dependence on technology, confidence, and cross-border connections.
Bailey says global steps toward safe AI model releases "should be a priority," The FSB is also studying how financial firms can safely use frontier models for cyber defense.
The letter therefore connects two different AI risks. One is financial: expensive AI assets, leverage, concentration, and cross-investment could amplify a correction. The other is operational: frontier AI could change how cyberattacks are carried out and how quickly they spread through connected institutions.
The broader debate over an AI correction
Bailey’s warning lands during a wider debate about AI investment stability. NYU finance professor Aswath Damodaran recently warned that an AI crash could hurt more than the dot-com bust. His point, as described in the source, is that the AI industry requires massive spending on physical infrastructure, and much of that spending is debt-financed.
That matters because the consequences of a correction may not stop with public-market investors. If debt-financed infrastructure is central to the boom, then a reversal could ripple through the broader economy.
For financial watchdogs, the question is not whether AI will matter. Bailey’s letter assumes it already does. The question is whether markets, regulators, and institutions are prepared for the risks that come with a fast-moving technology boom built on high expectations, borrowed money, concentrated providers, and frontier models that could reshape cyber threats.