Artificial intelligence was supposed to make the financial world faster, smarter and more efficient. But according to Bank of England Governor Andrew Bailey, there is a more uncomfortable question now demanding an answer:
What happens if the machines become too complex for humans to control?
Bailey has warned that as frontier AI systems grow more sophisticated, authorities must preserve the “right to intervene” when these technologies begin creating serious risks. He argues that the issue is not about stopping AI’s progress, but ensuring that society never loses the ability to step in when something goes badly wrong.
The warning comes against a backdrop of increasingly sophisticated cyber threats. AI could give attackers new capabilities capable of disrupting some of the basic machinery of modern finance — from everyday card payments and bank transfers to stock and bond trading.
For Bailey, this is no longer a distant science-fiction scenario. He described the risks from rapidly advancing AI models as “real and increasingly significant”, particularly as increasingly complex systems become harder to supervise.
Yet his message was not a call for an immediate regulatory crackdown.
Instead, Bailey argued that governments and financial authorities should first understand exactly how these systems behave. New AI models should undergo rigorous testing, allowing regulators to identify the points at which intervention would still be possible if a system began behaving dangerously.
His central principle is simple: innovation can move forward, but humans must retain the emergency brake.
And there is another warning emerging alongside the technological one — money.
The Bank of England’s Financial Policy Committee says the financial system is becoming increasingly exposed to the AI boom through a rapidly expanding mountain of debt.
Between January and September this year, major AI companies and related businesses raised around $450bn (£339bn) in debt. That figure has already exceeded the $333bn in UK government gilts scheduled for issuance across the whole of 2026.
The significance goes beyond the technology companies themselves.
Hedge funds, asset managers and private-credit firms are increasingly financially connected to an industry whose companies have yet to demonstrate widespread profitability. If expectations surrounding AI were to change sharply, those connections could transmit losses through broader financial markets.
The Bank’s committee described the rapid expansion of AI-related borrowing as broadening capital markets’ exposure to developments in the sector and stressed the need for careful management of the risks.
That leaves the world facing two intertwined AI questions.
Can humanity control increasingly powerful machines — and can the financial system withstand the enormous economic expectations being built around them?
For Bailey, the answer begins not with stopping AI, but with making sure that when the moment comes, someone still has the power to press pause.
