UK Financial Services AI Adoption Plan

Overview

The Financial Services AI Adoption Plan is an independent report published by HM Treasury on 14 July 2026. It sets out ten recommendations to accelerate safe and responsible artificial intelligence adoption across UK financial services. Its vision is for firms of all sizes to integrate AI into core processes at pace, moving beyond isolated pilots while maintaining trust. Intended benefits include better customer outcomes, productivity, sustainable economic growth, competitiveness and financial resilience. The plan emphasises coordinated action by industry, government and regulators rather than a separate statutory AI regime.

The report identifies fraud detection, operational streamlining, improved decision-making, personalised customer support and more responsive services as opportunities. The HM Treasury publication page states that government welcomes the plan and accepts recommendations addressed to government. Its adopted status reflects this policy acceptance, not legislative enactment or parliamentary approval. The report is non-binding: publication does not create new firm-level duties, penalties or a statutory entry-into-force date. Government acceptance also differs from completion of the recommended actions.

Definitions and Scope

AI is addressed through financial-services applications, including fraud detection, risk management, customer support and emerging payment applications. Generative AI features in the discussion of practical guidance on applying existing regulatory principles to particular use cases. The plan addresses adoption across firms of different sizes, including smaller firms and new entrants, and considers the interaction between regulated financial institutions and general-purpose AI tools. Its technological terminology serves a policy purpose rather than establishing a separate legal classification system.

The regulatory perimeter is the boundary between activities subject to financial regulation and activities outside it. The report concentrates on financial advice and guidance, including outputs from general-purpose AI tools that resemble personalised advice. The advice gap describes limited access to or use of regulated advice; the report states that regulated advice reaches only around 9% of UK adults. Agentic payments are an emerging AI use case capable of changing how transactions are initiated and managed. Consumer Duty, explainability, accountability and model risk management are identified as areas for clearer practical articulation within existing regulation.

Governance and Institutional Framework

HM Treasury is the publishing authority. The independent Financial Services AI Champions developed the substantive recommendations through engagement across the sector and work with government and regulators. Harriet Rees is identified as Group Chief Information Officer at Starling Bank, and Dr Rohit Dhawan as Head of AI and Advanced Analytics at Lloyds Banking Group. Their role is advisory and directed toward identifying barriers to adoption and developing recommendations. Government, financial regulators and industry are the intended participants in taking the plan forward.

The report treats the existing regulatory framework as a foundation for adoption and records industry support for a technology-neutral, outcomes-focused approach. It identifies the FCA AI Lab, AI Live Testing, the Supercharged Sandbox and the AI Input Zone as existing engagement and support pathways. It highlights fragmented guidance, uneven visibility of support and access challenges for smaller firms and new entrants. A clear, authoritative single source of cross-regulator guidance is proposed to improve accessibility and coordination. This is a proposal for practical support within the existing institutional framework, not a new statutory AI authority.

Recommendation Areas

The ten recommendations are organised into five groupings. Recommendation 1 concerns the regulatory framework; recommendations 2–3 concern the regulatory perimeter; recommendations 4–6 concern resilience; recommendations 7–9 concern skills and talent; and recommendation 10 concerns agentic payments readiness. Regulatory clarity is a central theme. Firms seek practical explanations of how existing requirements apply to AI and generative AI, particularly Consumer Duty, model risk management, explainability and accountability. Clearer support pathways and more consistent access to testing and sandbox opportunities are also priorities.

The perimeter discussion examines the contrast between regulated institutions and general-purpose AI tools producing financial outputs outside regulated channels. Consumers may not understand whether the service they are using is regulated, with consequences for safeguards and formal recourse. The report identifies opportunities to combine AI-driven triage, nudges and guidance with human expertise for more complex needs. Resilience, skills and talent, and agentic payments readiness form additional recommendation areas. Broader government considerations also include AI sovereignty and resilience, linking adoption ambitions to strategic capability and dependable financial services.

Implementation Approach

Implementation is collaborative rather than a legislated compliance programme. Government welcomes the report, accepts recommendations addressed to government and intends to work with regulators and industry on next steps. This acceptance concerns government recommendations rather than every action assigned to every participant. The report calls for sustained coordination as technology, market practices and expectations evolve. Its regulatory-clarity approach prioritises making support more accessible and translating high-level regulatory requirements into practical explanations for financial-services AI use cases.

The plan supplies strategic direction rather than a new authorisation route or exemption. Firms adopting AI remain subject to the requirements applicable to their activities. Preparatory management actions include identifying the regulatory questions raised by a use case, distinguishing binding rules from policy recommendations, and considering available engagement pathways. Testing and sandbox participation are support options rather than obligations created by the report, and participation should not be equated with automatic compliance. The checklist below separates these practical considerations from the report's non-binding legal character.

Monitoring and Evaluation

The report uses adoption surveys and stakeholder engagement to describe AI use and barriers to wider deployment. It cites an early-2025 DSIT survey in which 21% of firms in the combined financial and real estate sectors reported AI adoption, compared with 16% across the economy. Separately, it cites approximately 75% adoption among firms surveyed by the FCA and Bank of England, drawing on findings published in 2024. The Bank of England and FCA survey publication provides the financial-sector survey context.

These percentages concern different populations and should not be treated as directly comparable sector-wide measurements. The combined financial and real estate figure has a different scope from the financial regulators' respondent sample. Neither percentage measures universal adoption or establishes the safety and quality of individual deployments. The report also states that further FCA and Bank of England survey work is underway to update the sector picture. This evidence-gathering supports assessment of adoption trends rather than creating a report-specific statutory reporting duty.

The plan's stated goals provide qualitative dimensions for assessing progress: wider responsible adoption, better consumer outcomes, productivity, resilience and more accessible regulatory support. Firms may use internal tracking of AI use cases, outcomes, unresolved regulatory questions and engagement with support programmes as voluntary management practices. Such practices remain distinct from obligations arising independently under applicable financial regulation. Continued coordination is part of the plan's policy approach, not a mandatory monitoring framework imposed by the report.

Legal Effect, Liability and Consumer Protection

The report does not itself create financial penalties, criminal offences, civil causes of action, suspension powers or appeal procedures. Government acceptance does not transform its recommendations into enforceable regulator rules. Regulated firms continue to operate within the legal framework governing their activities, and enforcement or liability questions depend on the particular firm, activity and conduct. The plan's recommendations should therefore be distinguished from existing statutory and regulatory obligations when assessing governance, deployment decisions and customer-facing applications.

The perimeter discussion highlights consumer-protection risks where individuals rely on general-purpose AI financial outputs without understanding that regulated safeguards and formal recourse may not apply. This asymmetry supports examination of the advice-versus-guidance boundary and proportionate guardrails. It does not mean that an unregulated AI interaction is exempt from all civil, consumer or other law. The report creates neither a compensation scheme nor a complaints channel. Available redress depends on the provider, the service or activity, and the applicable legal framework.

Relationship to Other Instruments and International Ambition

The plan promotes better practical application of the existing UK financial regulatory framework rather than introducing a standalone AI-specific regime. Consumer Duty, model risk management, explainability and accountability are topics on which firms seek clearer application to AI. Targeted Support and simplified advice models are discussed as initiatives addressing the advice gap. These policy relationships do not amount to amendments or incorporation of other instruments. The FCA's innovation initiatives complement the plan through engagement, testing and support for firms exploring AI applications.

Government will consider the Champions' work alongside the Mills Review. That review, by Sheldon Mills, the FCA's Executive Director for Consumers and Competition, examines how AI could reshape retail financial services by 2030 and beyond and makes priority recommendations to the FCA. The year 2030 is a horizon associated with the Mills Review, not an implementation deadline or expiry date for the adoption plan. The two workstreams are connected through financial-services AI policy and consumer outcomes.

The report presents the United Kingdom as positioned to lead safe and responsible AI adoption globally, drawing on digital adoption, experience managing complex risks and robust regulatory frameworks. This is an ambition for international leadership and competitiveness, not a treaty commitment or recognition of equivalent compliance with another jurisdiction's AI regime. AI sovereignty and resilience are also within the plan's broader policy scope. Cross-border firms must distinguish these strategic ambitions from the legal requirements applicable to their operations in each jurisdiction.

Implementation Timeline

Practical Adoption Checklist

Sources and References