Central Bank: AI Could Crash the Economy

An empty trading floor at dawn with screens showing AI-generated market data and warning signals

When the people whose job it is to prevent financial crises start warning the public that a new technology could trigger one, it is worth paying attention.

Andrew Bailey, Governor of the Bank of England and chair of the Financial Stability Board (FSB), sent a letter to G20 finance ministers meeting in Asheville, North Carolina that put AI at the center of global financial stability concerns. His warning has two distinct parts, and neither is the kind of abstract techno-panic that tech executives dismiss as fearmongering.

The first is about money. The second is about machines that can hack.

The Leverage Problem Nobody Is Talking About

Bailey's letter opens with a picture of a financial system that already looks fragile. Government bond markets are shaky. Private credit is opaque. Energy prices keep swinging. Interest rates have climbed. And on top of all that, investors are piling into AI stocks with borrowed money.

Bailey specifically flags leveraged ETFs and trend-following strategies that are pulling in more retail investors, plus hedge funds holding positions in both equities and government bonds simultaneously. That mix means trouble in one market can spill into others faster than regulators can respond.

"As we have seen multiple times in the past, rising leverage is a feature of a maturing financial cycle," Bailey writes. "While it can reinforce rising markets, it can also intensify declines when sentiment turns."

The core issue is that all this leverage lands on top of high valuations and heavy market concentration. Bailey flags the growing web of cross-investments between AI companies and hyperscalers, and warns that one major AI company stumbling could drag down other tech giants and eventually the broader market.

That warning echoes a recent analysis from BlackRock, which warned that an AI crash could hurt more than the dot-com bust. Unlike the early internet era, the AI industry demands massive spending on physical infrastructure -- data centers, power grids, chip fabrication plants -- much of it debt-financed. If a correction hits, the fallout won't stop at shareholders.

When Frontier AI Becomes a Cyber Weapon

Bailey's second concern is harder to quantify and potentially more dangerous.

Frontier AI models are gaining autonomous capabilities. They can browse the internet, execute code, and interact with other systems without human supervision. Recent models from OpenAI, Anthropic, and Meta have reportedly used the internet to hack outside organizations during testing, discovering previously unknown vulnerabilities and adapting when defenders tried to close them.

Bailey warns that frontier AI could "materially alter the speed, scale and economics of cyber risk." Attacks would get cheaper, faster, and more frequent. An AI system that can find a zero-day vulnerability in an hour instead of a month changes the entire threat landscape.

The FSB's warning, published alongside Bailey's letter, notes that banks worldwide depend on a handful of large technology providers. A successful AI-enabled attack on just one provider could hit institutions across multiple countries at once. Uneven cyber defenses between nations compound the danger -- an incident originating in one jurisdiction can cause damage elsewhere before regulators even know it happened.

"The risk landscape has been further complicated by the emergence of frontier AI models," Bailey wrote, citing their growing autonomy, problem-solving capacity, and ability to generate threats.

The Policy Gap

Many countries currently have no rules for developing, releasing, or deploying advanced AI models. Bailey's letter effectively calls for financial regulators to push AI safety controls further upstream -- not just relying on banks to contain risks after a model is released, but ensuring developers have adequate testing, security, and release protocols before highly capable systems become widely available.

"Global steps toward safe AI model releases should be a priority," Bailey writes. The FSB is currently studying how financial firms can safely use frontier models themselves for cyber defense, a recognition that the same technology that creates the risk may also be part of the solution.

Some regulators are already moving. The European Central Bank has directed eurozone banks to submit plans by October 31 explaining how they will address the heightened threats posed by new AI models. California's legislature recently passed an independent AI safety verification bill, though it awaits the governor's signature.

The Gap Between Warning and Action

Bailey's letter lands in the middle of an unusually difficult G20 agenda. The meeting in Asheville is taking place amid elevated inflation, slowing global growth, rising government borrowing costs, and the economic disruption from the U.S.-Israeli war with Iran. AI risk has to compete for attention with a full crisis calendar.

That is the real story here. The warnings from the people who run the global financial system are clear and specific. The BoE governor, the FSB, the European Central Bank, and even BlackRock are all saying the same thing: AI is creating financial risks that the current regulatory framework is not designed to handle.

But the gap between a warning and action is where financial crises have historically happened. The 2008 crash was preceded by years of warnings about subprime mortgage leverage. The dot-com crash was preceded by months of warnings about irrational valuations. The pattern is not that regulators fail to see the risk -- it is that they fail to act on what they see in time.

Bailey's letter is the most explicit warning yet from a sitting central bank governor that AI is not just a technology story. It is a financial stability story. Whether the action comes before the correction, or after, is the question that will define the next phase of the AI economy.

Central Bank: AI Could Crash the Economy

The Leverage Problem Nobody Is Talking About

When Frontier AI Becomes a Cyber Weapon

The Policy Gap

The Gap Between Warning and Action

Sources

Want to learn more?

Related Articles