AI agents are already common across UK financial services, but firms are largely keeping them away from consequential decisions. The Bank of England and Financial Conduct Authority’s 2026 survey says 80% of respondents have agents in production and a further 14% plan to deploy them within 12 months. Most current uses support decisions rather than execute them: only 2% of firms reported agents with broad autonomy.
The wider adoption picture is equally important. Eighty-six per cent of respondents said they use AI, up from 75% in 2024. Third-party systems represented 39% of reported AI implementations and 53% of high-materiality systems. The survey also found concentration in infrastructure providers: the top three cloud and compute providers accounted for 79% of reported relationships. These are survey responses, not a census of the sector, and the regulators caution that some detailed questions had smaller response bases.
For governance teams, the practical message is that human oversight has not disappeared; it has moved into a more complex supply chain. Firms now need evidence that agent recommendations can be challenged, vendor changes can be tested and rolled back, and critical dependencies can be understood before automation expands into approvals, payments or trading.
