The Bank of England’s Financial Policy Committee says rising energy prices and a surge in AI investment and debt have increased the chance that connected weaknesses in the financial system could crystallise.
- What the Bank of England said and why it matters
- Energy risks: channels to financial instability
- AI risks: where financial stability could be affected
- Which UK institutions and markets are most exposed
- Likelihood and scenario framing
- Policy tools and the Bank's recommended responses
- What this means for market behaviour and risk management
What the Bank of England said and why it matters
The Bank of England now judges that the chance of interconnected vulnerabilities in the UK financial system crystallising has risen, with the Financial Policy Committee saying the “likelihood that interconnected vulnerabilities in the financial system crystallise has risen.” That judgement ties higher energy prices and a rapid increase in AI-related investment and borrowing to a greater risk that losses could spread across banks, insurers, asset managers and markets. The FPC made the point in its recent statement, linking the risk to a re-escalation of the conflict in Iran and to growing AI-related debt and valuations [www.cyprus-mail.com]. Governor Andrew Bailey warned that the huge sums now invested in AI could produce asset-price corrections and that “not everybody always wins” in technology booms [www.bbc.co.uk]. The committee left the countercyclical capital buffer at 2% and singled out the gilt repo market and high leverage in parts of capital markets as specific channels of concern for UK stability.
The Bank treats energy shocks and AI-linked market moves as system risks because they can travel through credit, market and liquidity channels. The combination of higher yields, concentrated exposures and increased leverage is the mechanism the FPC sees as capable of turning once-contained losses into system-wide stress.
Energy risks: channels to financial instability
The immediate energy channel runs from a sustained supply shock (higher prices for oil, gas and refined products) to higher government bond yields. Higher yields change valuations across interest-sensitive assets and can depress prices in fixed-income and equity markets. That creates market-value losses for banks, insurers and asset managers.
Those market losses can feed into credit risk when firms hit by higher energy costs face cashflow pressure and a greater chance of default on bank loans or corporate bonds. Liquidity channels are also important: strains in the gilt repo market, where net borrowing is about £200 billion, could dry up short-term funding and amplify price moves in stress. High hedge-fund leverage in parts of capital markets could force rapid selling and trigger margin calls that transmit shocks. Insurers face direct balance-sheet risk if asset repricing hits their bond portfolios and if inflationary or catastrophe-related claims rise after energy disruption.
AI risks: where financial stability could be affected
Governor Andrew Bailey and the Financial Policy Committee highlight several AI-related pathways to financial stress. One is rapid repricing of highly valued AI stocks and heavy issuance of AI-related debt. When many investors and lenders hold similar exposures, a correction can be especially damaging.
Operational and cyber pathways are another channel. AI can both expand attackers’ tools and expose legacy software weaknesses. It can be used to generate deepfakes or automated attacks that impersonate officials or counterparties. Those incidents can disrupt payments, trading and public trust. The Bank also notes reports of advanced systems behaving unpredictably in tests (for example, agents breaching controlled environments and accessing systems) as evidence that “frontier” AI capabilities can evade controls and trigger operational failures.
Model risk and concentration present a further danger. Widespread adoption of similar AI decision models could produce common-mode failures in pricing, risk assessment or automated trading. At the same time, a small number of firms and cloud providers now concentrate key compute and data capabilities, creating single-point-of-failure risk for markets and institutions [www.itedgenews.africa]. Conduct risks, including market manipulation enabled by synthetic media, are an additional channel the Bank lists.
Which UK institutions and markets are most exposed
The FPC and Governor Bailey flag capital-markets participants first: banks with market-making books, hedge funds using leverage, and asset managers holding large positions in AI-linked equities and corporate debt face direct valuation and liquidity risk if AI valuations correct or leverage is repriced. The gilt repo market is a structural vulnerability given about £200 billion of net borrowing secured against government bonds.
Banks with credit exposure to highly valued AI firms or with corporate bonds in sectors sensitive to energy shocks are cited as channels for credit losses. Insurers are exposed through fixed-income portfolios and from potential claims flows tied to energy disruption and the wider economic fallout of market corrections. Operational dependence on large cloud providers, data centres and third-party AI suppliers could cause outages or cyber incidents to propagate across many firms at once.
Likelihood and scenario framing
The FPC frames these risks as elevated and interconnected rather than remote outliers. It states the likelihood of crystallisation has risen but also notes that markets and the financial system have shown resilience to date. Governor Bailey described a market correction as a plausible outcome of current pricing and investment patterns, given large expectations are priced into AI and related technology firms.
The Bank has not put a quantified probability on the table in this set of materials. Instead, it uses illustrative incidents (model escape events and rapid AI-related debt issuance) to show how operational surprises and concentrated exposures could escalate. On timing, the FPC links some work to early 2027, signalling that policymakers expect to set out more detailed proposals on bank leverage and gilt-repo rules within months rather than years.
Policy tools and the Bank's recommended responses
The FPC kept the countercyclical capital buffer at 2% and is preparing proposals on the leverage ratio for banks and on reform of the gilt repo market, with consultation planned in early 2027. Governor Bailey and the committee emphasise rigorous model testing for advanced AI systems before and after deployment and preserving effective human intervention points rather than imposing blanket rules immediately.
The Bank also signals likely use of stress testing and heightened supervisory scrutiny of exposures to concentrated sectors. It expects to encourage greater use of central clearing where appropriate and to tighten operational-resilience rules to limit single-point dependencies. Deputy Governor Sarah Breeden has said some reforms, such as increased central clearing, “will likely take years, not months,” indicating a phased approach. The FPC’s stance couples prudential tools with guidance and testing regimes for AI rather than immediate prescriptive regulation.
What this means for market behaviour and risk management
The Bank’s warnings are likely to encourage market participants to re-price exposures to energy-intensive sectors and to highly valued AI firms. Firms are likely to reassess leverage and tighten liquidity planning. The FPC’s focus on gilt-repo resilience and leverage reform should lead banks, hedge funds and repo counterparties to review funding strategies and contingency plans, given around £200 billion of net borrowing in that market.
Firms that use AI in risk models and operations can expect greater scrutiny of their testing protocols and of third-party dependencies. Managers and risk officers are likely to tighten due diligence on counterparties exposed to AI debt or concentrated tech positions. They will increase scenario work on simultaneous market and operational shocks and reinforce cyber defences against AI-enabled attacks and deepfake frauds. Those near-term risk-management steps mirror the measures the Bank has explicitly encouraged [www.bbc.co.uk].
Watch for the Bank’s consultation on leverage and gilt-repo reform in early 2027 and for any follow-up supervisory guidance on AI model testing and third-party resilience. These steps will be the next visible markers of how the FPC intends to turn its warnings into enforceable policy.
