Skip to main content

FPC September record: interconnected risks sharpen, AI debt moves centre stage

The Bank of England's Financial Policy Committee has warned that financial stability risks have worsened since July, with sovereign yields at post-2008 highs and AI-related debt issuance broadening capital market exposures. Senior leaders in banking, asset management and insurance now face a stability signal that reframes capital planning, counterparty exposure and operational resilience priorities.

The Financial Policy Committee's September record is the clearest stability warning the Bank of England has issued this year. Interconnected vulnerabilities are more likely to crystallise simultaneously, sovereign bond yields across advanced economies have reached levels not seen since 2008, and AI-related debt issuance is pulling capital markets further into a concentrated exposure (Bank of England). For boards and executive committees, the signal is not a single risk to price but a set of correlated risks to plan against.

Key Executive Takeaways

  • The Bank of England's FPC has judged that financial stability risks worsened between July and September 2026, driven by Middle East conflict re-escalation, higher sovereign yields and rapid growth in AI-linked debt (Bank of England).
  • Senior leaders should assume correlated shocks across sovereign debt, risky credit and AI-exposed equity rather than treating them as independent exposures.
  • Operational resilience planning must now explicitly address frontier AI cyber and model-access risks, which the Governor flagged to Court in July (Bank of England).

The correlation problem

The FPC's framing matters more than any single number. The Committee has moved from describing discrete vulnerabilities to warning that sovereign debt stress, stretched risky asset valuations and risky credit markets could adjust together (Bank of England). That changes the arithmetic of stress testing. Scenarios that assume diversification benefits across asset classes will understate loss paths if the same macro trigger, in this case a protracted negative supply shock from oil and gas prices, drives yields wider and valuations lower at once.

Hedge fund leverage in the gilt market remains elevated, even if stable, and the Committee has again pointed to its work on gilt repo market resilience (Bank of England). For asset managers and bank treasuries, that is a direct prompt to review repo counterparty concentration and margin assumptions before any sharp adjustment forces the issue.

AI as a balance sheet exposure

The more novel element is AI. The FPC notes that AI-related debt issuance has risen rapidly, broadening capital markets' exposure to developments in the sector, and that equity valuations for AI companies fell sharply in July with the adjustment amplified by an unwinding of stretched positions (Bank of England). This is no longer a thematic equity story. Credit portfolios, private debt funds and insurance investment books now carry AI exposure through fixed income instruments that were not material a year ago.

The operational dimension compounds the financial one. In July, Governor Bailey told Court that the Bank was supporting banks to ensure sufficient access to AI models to protect cyber security (Bank of England). The FPC has since reinforced that recent incidents in frontier AI have drawn focus to cyber and operational resilience risks (Bank of England). Firms that treat AI governance as a model risk question alone are missing the resilience and third-party dimensions supervisors are now watching.

Board positioning

The FPC's judgement that UK households, businesses and banks remain resilient is a floor, not a ceiling. The Committee's view that the banking system is strong enough to support borrowers in a stress (Bank of England) gives executives room to extend credit through volatility, but only if their own capital and liquidity planning reflects correlated downside paths.

For chairs and risk committees, the practical response is to re-run stress scenarios with sovereign yield, credit spread and AI-exposed equity shocks moving together, and to ask whether operational resilience frameworks have caught up with how quickly AI exposure has entered the balance sheet. The FPC has done the diagnostic work. The positioning decisions sit with boards.

What this reveals

The FPC's warning exposes a widespread planning assumption that risks can be modelled as discrete exposures with diversification benefits intact. When a single macro trigger drives sovereign yields, risky credit and AI-exposed equity in the same direction, stress tests built on historical correlations understate loss paths and capital planning carries hidden fragility. Many boards will assume their counterparty, repo and credit books are independently sized when they are in fact exposed to the same underlying shock, and operational resilience plans drafted before AI became a balance sheet exposure will not reflect the frontier cyber and model-access risks the Governor has now flagged. The broader issue is that stability signals from supervisors are moving faster than internal risk frameworks are being updated.

Questions accountable leaders should ask

  • 01When did your stress scenarios last assume simultaneous adjustment across sovereign debt, risky credit and equity, rather than treating these as independent shocks?
  • 02Can you evidence, at counterparty level, your exposure to AI-related debt issuance across credit portfolios, private debt holdings and structured positions, not just thematic equity?
  • 03Does your repo and margin planning reflect current hedge fund leverage in the gilt market, or assumptions set before the FPC's recent interventions?
  • 04Has your operational resilience framework been updated to address frontier AI cyber and model-access risks, or does it still treat AI as a productivity theme?
  • 05If the FPC's correlated-shock scenario crystallised next quarter, could your board defend the capital, liquidity and counterparty decisions it approved this year as having tested that possibility?

What accountable leaders should do now

  1. 1Commission an immediate review of stress testing assumptions to confirm scenarios include a single macro trigger driving sovereign yields, credit spreads and AI-exposed equity to adjust together, and quantify the loss path gap against current assumptions.
  2. 2Map AI-related exposures across the full balance sheet (public credit, private debt, structured products, counterparty concentrations) and report concentration metrics to the risk committee before year-end planning concludes.
  3. 3Pressure-test repo counterparty concentration and margin assumptions against the FPC's gilt market resilience work, and identify where a sharp adjustment would force operational decisions you have not yet rehearsed.
  4. 4Extend the operational resilience self-assessment to explicitly cover frontier AI cyber risk and third-party model-access dependencies, and brief the board on the gap between the current framework and the Governor's July signal.
  5. 5Record, in the board minutes of the next risk discussion, which FPC signals were considered, which were discounted and why, so the reasoning behind this quarter's capital and resilience decisions is defensible later.

Explore the practical guide

This guide sets out how to pressure-test a stakeholder map before committing to a regulated market launch. After reading, you will know which gaps matter, how to find them, and when your map is complete enough to act on.

Read the guide

Where the operating environment may be moving faster than internal reporting reflects

Polar Insight's Signal Briefings translate emerging regulatory, stakeholder, and market developments into a clear implication for accountable leaders.

Explore Signal Briefings

Stakeholder Signals

Consequential developments in financial services and other regulated markets, with one implication for accountable leaders.