Bank of England

The Bank of England's engagement with AI risk deepened in April 2026 when its Financial Policy Committee issued a formal record directing the BOE and the Financial Conduct Authority to Conduct further work on agentic AI in payments and financial markets, citing the systemic risk as 'likely to increase rapidly' given that 75% of UK financial firms are already deploying AI. The Governor committed to the Treasury Committee to Conduct biennial AI-specific surveys of UK financial sector AI adoption.

The BOE's employment modelling has taken on new urgency. The Financial Policy Committee used 500,000 extra unemployed as its AI displacement worst case, anchored to a vacancy base above 730,000. As of the ONS bulletin published 18 June 2026, UK vacancies stood at 707,000, below that baseline. The threshold has been crossed without the mechanism being named: the ONS still publishes no AI-attribution layer, and payrolled employment is down 138,000 year on year. The BOE's worst-case model is now technically in-territory while its measurement infrastructure cannot confirm why.

This is a distinct risk framing from the BOE's earlier warnings about AI-sector asset overvaluation: the FPC directive focuses on operational stability in payments infrastructure, while the vacancy breach sits in the employment mandate. The BOE now spans two independent AI risk mandates: the market-stability lens (AI-sector overvaluation in equities held by UK insurers and pension funds) and the operational-stability lens (agentic AI in settlement chains). The Treasury Committee's call for AI-specific stress tests signals that voluntary BOE guidance is hardening into formal supervisory expectations, while the labour data it needs to validate its own models remains unattributed.