Andrew Bailey does not usually write personal essays. But the governor of the Bank of England just used his first piece for the bank's new "Insight" series to make a specific argument: society must never lose the ability to step in and stop an AI system, no matter how good the technology gets.
His words matter more than most warnings about AI, because Bailey is not a random commentator. He runs the central bank that manages the pound, oversees British banks, and as chair of the Financial Stability Board, coordinates financial rules across the entire world's major economies.
He has been saying versions of this for months. In July, at his Mansion House speech, he pushed back on calls to loosen AI oversight in finance. In August, writing to G20 finance ministers, he flagged that AI could speed up cyberattacks faster than banks can defend against them. This week's essay is the clearest version yet: keep the option to intervene now, figure out the rules later.
That second part is the interesting bit. Bailey is not asking for new AI laws today. He thinks regulators do not yet understand these systems well enough to write good rules, and rushing into "regulatory architecture" before that happens would be a mistake. His starting point instead is rigorous testing of AI models, so authorities can find the exact places where they would need to step in if something broke.
The essay landed on the same day the Bank's financial policy committee released a number that explains the urgency. Large AI companies took on 450 billion dollars in debt between January and September this year. To put that in perspective, the UK government's entire bond program for next year is smaller, at 333 billion dollars. Nine months of AI company borrowing already beat a full year of UK government borrowing.
Who is holding that debt matters as much as the size of it. It is not just venture capital funds anymore. Hedge funds, asset managers, and private credit firms have all taken positions tied to AI companies that, by and large, are still not profitable. If a major AI company stumbles, the pain does not stay inside tech. It runs straight into the same funds that manage pensions, insurance reserves, and corporate cash.
This connects to a pattern other institutions have already flagged. The Bank for International Settlements warned in its June report that AI spending has become a genuine financial stability risk. The IMF has separately warned that AI has turned cyber risk into a systemic threat to the financial system, not just an IT problem for individual companies.
None of this means new rules are coming next month. But it does mean the world's central bankers have stopped treating AI as a side issue for tech regulators and started treating it as a mainstream risk to markets, credit, and daily banking. For anyone running a business, that shift in tone from the people who set interest rates and back up the banking system is worth watching closely, because when central banks start talking about "the right to intervene," it usually means they are preparing for a moment when they will need to use it.