The Review also outlines seven recommendations for the FCA Board and Executive to consider:

1. Secure and adapt the regulatory perimeter

The Review recommends that the FCA should consider securing and adapting the regulatory perimeter for AI-mediated retail financial services by launching a review into the scale, nature and impact of general purpose Large Language Models (LLMs) outside the perimeter.

The Review says that this review should be launched as soon as possible and finalised within the next 3-6 months – suggesting there is a degree of urgency in this space. The intention is that the FCA can determine whether to ‘amend guidance, recommend perimeter changes to government, or maintain the current approach’.

Among the issues for the review would be how advice and guidance boundaries operate, how by way of business tests apply to general purpose tools, and how financial promotions and arranging rules apply. This is a matter that has been of concern to some firms who perceive that LLMs are undertaking activities which ought properly to be within the regulatory perimeter. This relates not only to investment advice but also savings, debt management, pensions and drawdowns, mortgages and home loans.

In the longer term, given the risk of circumvention/gaps in the regulatory framework, the Review says the FCA should consider requesting the government to boost the FCA’s existing powers under the Critical Third Parties regime (currently restricted to systemic issues) and the Designated Activities Regime (DAR) (where HMT can bring specific activities within the perimeter on the FCA’s recommendation) to meet these challenges.

The Review also suggests that direct powers under the Digital Markets, Competition and Consumer Act (DMCCA) should be given to the FCA and other sectoral regulators (such as the Competition and Markets Authority).

2. Strengthen system-wide coordination and oversight

The Review recommends that the FCA consider enhancing and establishing coordination across both domestic authorities and international partners to address the risks and opportunities of AI. In the absence of a dedicated AI regulator or a dedicated AI cross-sector regime including legislation, coordination is an imperative for UK sectoral regulators.

3. Monitor the transition to autonomous models and adapt regulatory frameworks

The FCA should consider monitoring the transition towards more autonomous AI and adapt its regulatory frameworks accordingly. The FCA, alongside the PRA, should support the development of more effective approaches to AI model risk management and should work with firms, model providers and researchers to understand how advances in AI can improve explainability, assurance, governance and oversight.

4. Scale up the FCA's AI Lab to support AI models and system innovation in financial services

The FCA should consider establishing a structured capability, anchored in the AI Lab that it launched in 2024, to assess AI models and systems used in financial services. This would include working with firms, model developers, researchers and technical experts to help the FCA anticipate how new model capabilities and system designs could shape financial services before they become embedded in core firm operations. This would help the FCA develop an independent understanding of AI models and systems, use its convening role to support responsible growth, and address challenges around explainability, assurance and governance in financial services AI.

5. Enable the foundations for agentic finance

The FCA should consider leading the development of a trusted framework for AI agent participation in financial services, clarifying how agents can be authorised, identified and held accountable. This should help establish clear expectations for consent mandates, identity, control and liability, creating the conditions for safe adoption of more autonomous AI-enabled services.

6. Build and adopt an AI-enabled Agentic Supervisory Model

See the comments above regarding the need for supervisors to use AI to identify cross-firm patterns, emerging harms and system-wide risks. The Review calls this the Agentic Supervisory Model.

The Review recommends that the FCA consider developing an AI-enabled agentic supervisory model that enhances both firm-by-firm regulatory efficiency and system wide oversight. This could fundamentally change the nature of compliance, supervision and enforcement, increasing the ability of supervisors to detect regulatory breaches, identify weak governance or controls, and intervene to prevent system-wide harm.

7. Develop a trusted public-interest AI-enabled financial capability service

The FCA should consider taking a more proactive, ‘hands on’ approach to financial capability and access to future vital AI-enabled services.

The FCA should also consider convening the development of a free, inclusively designed, trusted AI-enabled financial capability and support system that has a range of digital tools that provides consumers with access to reliable financial information, guidance and support from trusted sources.