The Bank of England's Financial Policy Committee has set out plans to scrap a longstanding buffer within the leverage ratio framework, a change that would primarily benefit the largest UK domestic-focused lenders — NatWest, Lloyds, Nationwide and Santander UK — according to the record of the FPC's 26 June 2026 meeting, published 7 July The Guardian.

The proposed changes could reduce the leverage ratio of the affected lenders by 20 basis points on average, the FPC said. The leverage ratio measures a bank's core capital against its total, unweighted exposures, and functions as a backstop to risk-weighted capital requirements — a floor beneath the more complex, model-driven measures that determine how much capital banks must hold against different classes of assets. Trimming the buffer built into that ratio effectively frees up capital that lenders could otherwise deploy toward lending.

The FPC's own record is explicit about the rationale for the buffer it now proposes to remove: capital buffers exist to be drawn down and support lending during periods of stress, not to sit permanently unused Bank of England. The Committee's language frames the leverage ratio review as part of a broader recalibration of the post-2008 capital framework, a theme picked up in a companion "Financial Stability in Focus" piece published the same day examining the bank capital framework more broadly, alongside the July 2026 Financial Stability Report Bank of England.

The move is not without internal dissent. Some FPC members raised concerns during the meeting that loosening capital requirements at the banking-sector level could inadvertently encourage a build-up of leverage elsewhere in the financial system — specifically, debt-fuelled positions taken by hedge funds in AI-related equities The Guardian. That concern sits alongside a separate, and in some ways more striking, warning in the same record: that rapid advances in frontier AI capabilities have themselves become a source of financial stability risk, particularly around cyber and operational resilience.

The FPC noted that AI development has outpaced some earlier expert forecasts, a dynamic it says lowers the cost and raises the scale at which malicious actors can inflict shocks or outages on financial infrastructure. Notably, the Committee's discussion referenced Anthropic's Mythos model — a system that has, to date, only been made available to a small number of vetted companies globally — as an illustration of how quickly frontier capability is advancing relative to the industry's ability to assess and contain the risks it introduces The Guardian.

Taken together, the record reads as a Committee trying to hold two positions at once: easing prudential constraints on the core banking sector while simultaneously flagging that the frontier of financial-system risk is migrating toward leveraged non-bank actors and AI-enabled operational threats. That juxtaposition is worth sitting with. Capital buffers were designed, in the post-crisis architecture the FPC itself invokes, to be the shock absorber of last resort for the regulated banking sector. Relaxing them on the premise that they're not being used as intended, while in the same breath warning that leverage is building up in less-regulated corners of the market chasing AI valuations, suggests a regulator managing two different clocks — one calibrated to bank balance sheets, the other to a faster-moving technological and market cycle it does not yet fully control.

The FPC has now launched a review to identify financial stability gaps arising from the capital rule changes, with that work due to conclude by the end of September 2026. Its findings are intended to feed into a formal consultation package expected in early 2027 Bank of England. That timeline gives the Bank roughly a quarter to assess second-order effects — including the market-leverage concerns some members raised — before committing to a rulemaking process that could stretch well into next year.

The institutional backdrop adds another layer. The Prudential Regulation Authority's Annual Report for 2025/26 confirms that Katharine Braddick's term at the PRA began on 1 July 2026, placing new leadership at the supervisory body just days before the FPC record's publication Bank of England PRA Annual Report. The Bank's stress-testing page was also updated as recently as 3 July, suggesting the scenario framework used to evaluate bank resilience is being refreshed in parallel with the capital-rule review — though the record does not detail whether AI-related risks, cyber shocks among them, will feature explicitly in future stress scenarios.

For banks, the near-term effect is straightforward: more usable capital, and potentially more headroom for lending growth, once the rule change is implemented. For the FPC, the harder question is whether the September review will conclude that easing bank-level buffers is safe only if matched by tighter oversight of leverage building up outside the regulated banking perimeter — in hedge funds, in AI-exposed portfolios, and in the operational plumbing that increasingly capable AI systems are beginning to touch.