The Rapid Rise of AI in UK Finance: Key Stats, Risks, and the Push for Sovereignty

• Widespread Adoption: 86% of surveyed financial firms currently utilize AI, with generative AI accounting for 56% of all reported systems in 2026.
• Agentic Growth: While 80% of respondents have deployed AI agents, only 2% grant them broad autonomy, favoring decision support and constrained execution.
• Cloud Concentration: There is extreme reliance on third-party infrastructure, with the top three cloud providers handling 79% of all reported relationships.
• Risk Priorities: Cyber risk amplification is the primary concern for 57% of firms now, a figure projected to rise to 80% within three years.
• Government Intervention: The UK government has injected an additional £60m into the Sovereign AI Fund to bolster domestic capabilities and reduce foreign dependence.

The survey, titled ‘Artificial intelligence in UK financial services - 2026’ and conducted jointly by the BoE and the Financial Conduct Authority, indicates that adoption is nearly universal among major institutions. All responding UK banks and international banks reported current AI usage, followed closely by insurers at 93%. When looking at the broader pipeline of implementation, the numbers climb even higher: 94% of respondents have at least one AI system in production. This figure rises to 97% when including systems currently in pilot or proof-of-concept stages, and reaches 99% when accounting for planned deployments over the next twelve months.

Despite this near-total penetration, the distribution of AI use is not uniform across all business areas. Usage remains heavily concentrated in operations and IT, which together account for 31% of all reported systems. James Maxfield, Chief Client Officer at the London-based tech firm Duco, notes that operations is one of the fastest-growing areas for AI integration. He points out that seven in ten firms are either already using AI in this sector or planning to do so within the next year, presenting a significant opportunity for post-trade teams and operational efficiency.

A defining feature of the current wave of adoption is the proliferation of AI agents. The BoE report highlights that 80% of respondents have deployed AI agents, with an additional 14% planning to do so within the next year. However, the level of autonomy granted to these agents remains strictly controlled. Only 2% of respondents reported deploying agents with broad autonomy.

The majority of deployments fall into more restrictive categories: 80% are used for decision support, 46% have constrained execution capabilities, and 22% operate with conditional autonomy. When agents do possess higher levels of autonomy, they are most likely to be found in Operations and IT, Risk and Compliance, and Sales and Marketing departments.

Maxfield observes that while firms are rigorous about approving AI systems before they go live, ongoing monitoring is less consistent. He suggests that only around half of the firms keep strict post-deployment oversight, arguing that continuous assurance comes from the people who understand the underlying work best.

Financial institutions report deriving their greatest current benefits from operational efficiency, automation, and productivity gains. These are followed by improvements in cybersecurity, operational resilience, and risk management control effectiveness. However, these benefits come with substantial risks.

Cyber risk amplification is the most significant concern for the industry. Currently, 57% of respondents rate this risk as high or severe. This anxiety is expected to intensify, with that figure projected to jump to 80% within the next one to three years. Similarly, concerns about critical third-party dependencies are rising; the share of firms rating this concern as high or severe has increased from 40% today to 63% for the near future.

This dependency is starkly illustrated by the concentration of cloud infrastructure. The report reveals a heavy reliance on the three largest cloud providers, which account for 79% of all reported cloud and compute relationships. In fact, 98% of respondents use at least one of these top three providers. Of the 169 distinct cloud and compute relationships reported, 37% were deemed both highly critical and difficult to substitute. Alarmingly, only 35% of that high-criticality group had tested exit plans, leaving many firms vulnerable to potential disruptions or vendor lock-in.

In response to these vulnerabilities, there is a growing push across the UK and Europe to reduce dependence on US tech giants. Research from Swiss firm Proton indicates that more than 74% of publicly listed European companies rely on US-based tech services. The BoE’s findings mirror this trend, highlighting the dominance of presumed US firms like Amazon, Microsoft, and Google in the UK financial sector.

To counter this, the UK government is taking active steps to bolster its position in the global AI race, which is currently dominated by the US and China. Prime Minister Andy Burnham has appointed Kanishka Narayan as the UK's first AI Minister. Recently, the government announced an additional £60m (US$80m) for the Sovereign AI Fund, building upon a previous £500m (US$671m) investment announced under former Prime Minister Keir Starmer. This new funding aims to enhance direct investment in promising UK AI companies.

Further support includes an additional £60m (US$80m) for the Sovereign AI’s R&D Procurement Scheme and £18m (US$24m) for a new UK Advanced Chip Design Academy, addressing Britain’s heavy reliance on semiconductor imports.

Leading financial institutions are beginning to act on these concerns. Starling Bank has become the first UK bank to train an AI model on UK sovereign cloud infrastructure. In partnership with NVIDIA, Starling developed its TFM model using Civo's sovereign infrastructure, which will underpin its UK-first agentic AI assistant.

Harriet Rees, Group CIO of Starling Bank and a UK Government AI Champion for Financial Services, emphasizes the strategic importance of this shift. She argues that the UK needs more sovereign compute and cloud capacity, urging British businesses to support domestic infrastructure. Without such measures, she warns, the UK risks losing its status as a leading global AI ecosystem and a leader in wider industries. As the sector continues to evolve, balancing rapid innovation with robust, sovereign infrastructure will remain the central challenge for UK financial services.