The Bank of England (BoE) warned Wednesday that rapid increases in the level of AI-related debt, combined with recent security concerns around the technology, are posing a risk to UK financial markets’ stability.

In a record of the Financial Policy Committee meeting on 25 September, the central bank said that the growing indebtedness of AI firms and their opaque and, at times, “circular” financing arrangements could complicate risk analysis and potentially amplify losses if they are unable to deliver.

AI-related investment has continued to grow rapidly since the Committee’s last meeting, it added, increasing the extent to which AI developments could affect “a wide range of investors and funding markets”. Global debt issuance for AI-related investment is expected to exceed that of countries such as the UK in 2026, with JPMorgan estimating that around $450 billion of debt has already been supplied this year.

Although sterling-denominated AI debt is “significantly smaller” than in the US or eurozone, it still accounted for 47 per cent of GBP corporate bond issuance this year.

Alongside this, private credit is expected to play an increasingly important role in this financing, with Morgan Stanley analysts estimating that $700 billion of data centre capital expenditure between 2026 and 2028 will be funded by private credit.

Private credit markets are relatively opaque, and their rapid growth has triggered bodies including the Financial Stability Board to warn that the “complex interlinkages” between funds and traditional banks pose a threat to the global financial system.

The BoE also noted that the large amounts of debt are predicated “in part” on the expectation that AI will generate significant productivity gains. If these fail to materialise, therefore, it could affect “not only AI-related asset valuations but also sovereign debt markets”.

Outside of their credit risk, the bank warned that AI models are increasing the threat of cyber and operational risks to businesses.

Citing recent high-profile hacks conducted by frontier labs’ models during testing, the BoE said that “containment, monitoring and governance arrangements” could be challenged further as their capabilities increase.

Malicious actors using open-weight models, which have more easily modifiable safeguards, could also pose a growing threat to financial stability, it said. The increased pace and scale of vulnerability identification has made vulnerability patching “critically important”, the bank added, though it believes these models could also bring benefits to firms’ cyber defences.

AI was not the only risk noted in the meeting.

The “re-escalation” of conflict in the Middle East, including the ongoing war between the US and Iran, is putting pressure on oil and gas prices, the bank said. At the same time, pressure on refineries has increased the price gap between Brent crude oil prices and those of refined petroleum products.

The “more protracted” negative supply shock has renewed uncertainty around growth and interest rates in several advanced economies, which the BoE said could “expose existing financial vulnerabilities, amplify stress and impair the provision of vital financial services.”

Risky credit markets, including parts of private credit, are especially vulnerable to any tightening in financial conditions, it added, noting that risk taking in these areas “remained elevated”.

---