Independent report

Financial Services AI Adoption Plan

Published 14 July 2026

Vision 

Artificial Intelligence (AI) is already reshaping UK financial services, improving fraud detection, streamlining core operations and sharpening risk management. The opportunity now lies in moving beyond isolated pilots to rapidly scaled AI across the entire sector to deliver tangible benefits for consumers, businesses and the wider economy.   

With high levels of digital adoption and world-class regulation, the UK financial services sector starts from a position of strength. Its experience in complex risks under robust regulatory frameworks makes it uniquely positioned to lead safe, responsible AI adoption globally. This is not an incremental opportunity; scaling AI is a strategic imperative for UK growth, competitiveness, and financial resilience.      

Our vision is a financial services sector where firms of all sizes confidently integrate AI across core processes responsibly and at pace, leading to better customer outcomes, increased productivity, and sustainable growth for the UK economy. Realising this vision will require a strong, collective focus on scaling AI as a strategic priority, while maintaining trust and ensuring the resilience of the financial system. As technologies evolve and expectations change, sustained progress will depend on effective and continued coordination between industry, government and regulators.  

Through this shared effort, the UK can unlock safe, widespread adoption, improve outcomes for consumers and businesses, and strengthen its position as a global leader in AI‑enabled financial services.

Context 

The financial services sector is an early adopter of AI and stands among the best-positioned industries to realise immediate productivity gains. DSIT’s AI Adoption Survey from early 2025 showed 21% of firms in the financial and real estate sectors had already adopted AI, above the 16% seen across the economy as a whole.^{[footnote 1]} Across the economy, initial adoption was driven by larger firms, where adoption rates were over double those seen in smaller firms. This is reflected within the financial services sector, with adoption rates reported by firms surveyed by the Financial Conduct Authority (FCA) and Bank of England (BOE) in around 75%, significantly higher than the average across the economy^{[footnote 2]}. This figure reflects findings published in 2024, with further survey work currently underway by the FCA and Bank of England to provide an updated view of adoption across the sector.       

AI presents a clear opportunity to deliver tangible benefits across the sector. For consumers, this includes more effective fraud detection, personalised and accessible financial support, and faster, more responsive services. For businesses, AI has the potential to significantly improve productivity by streamlining operations, enhancing decision-making, and enabling new products and services. Together, these use cases represent a significant opportunity to improve customer outcomes, increase productivity, and support UK economic growth. 

Emerging use cases, such as agentic payments, illustrate how AI could transform how transactions are initiated and managed, supporting more seamless and efficient commerce. While adoption has progressed rapidly, this represents an early stage in the development of AI across financial services, with significant further opportunities emerging as technological capabilities evolve and deployment continues to scale. 

Unlocking the full potential of AI across financial services will require addressing the key barriers identified across the sector. Our stakeholder engagement across industry, regulators, and trade bodies revealed five key themes where action can significantly accelerate safe adoption. These are areas requiring greater clarity, coordination, or investment rather than insurmountable barriers, meaning that with the right support, UK financial services can continue to innovate confidently and at pace to deliver improved outcomes for consumers.

Key Themes and Issues

1. Regulatory Clarity

The UK’s existing regulatory framework is widely seen as a major asset and a strong foundation for AI adoption, with firms strongly supporting the regulators’ tech-neutral, outcomes-focused approach over the introduction of new AI-specific regimes. Regulators have already launched a wealth of highly valuable initiatives, guidance and innovation pathways designed to support the sector, which have been welcomed by industry. This includes the Financial Conduct Authority’s (FCA) AI Lab, which offers initiatives such as AI Live Testing and the Supercharged Sandbox, enabling firms to engage with the FCA on novel AI use cases. It also includes the AI Input Zone, which has recently sought feedback on examples of good and bad practice to inform a planned FCA publication later this year. 

The core challenge now is not the absence of regulatory support, but its accessibility, consistency and practical application across the full breadth of the sector. 

Firms frequently reported access to participate with sandboxes is limited, and that relevant guidance is fragmented across multiple sources, making it difficult to navigate the regulatory landscape more broadly. There is limited visibility on how to access existing regulatory support, meaning the full impact of innovation pathways, is not yet being felt universally. Furthermore, despite the regulators’ foundational work, firms continue to request more practical articulation of how high-level principles apply to AI and generative AI use cases, particularly regarding the Consumer Duty, model risk management, explainability and accountability. 

Engagement with regulators is viewed positively, particularly when discussing engagement on real use cases. The ambition now must be to broaden access; currently, engagement and sandbox usage can be inconsistent, with smaller firms and new entrants finding these pathways harder to navigate.       

To ensure AI adoption does not stall or become uneven, the next phase of regulatory policy must focus on scaling the reach of these existing successes. The priority now should be to establish a clear, authoritative single source of cross-regulator guidance, enabling firms to navigate requirements confidently and scale adoption consistently across the sector. Where appropriate, we are supportive of the regulators using AI to enhance the delivery of support services.

2. Regulatory Perimeter

The UK continues to face a persistent and structural financial “advice gap”, with regulated advice currently reaching only around 9% of UK adults.^{[footnote 3]} The government and regulators have responded with a range of initiatives, including Targeted Support, simplified advice models, as well as industry initiatives including public awareness campaigns aimed at improving engagement. However, while directionally positive, these have not yet shifted the fundamental economics or materially closed the gap at scale. 

Today, consumers are routinely turning to general-purpose AI tools for everyday budgeting, saving and investment tips, yet they cannot access equivalent features from their own banks. Consumers are turning to these unregulated spaces often unaware these are outside regulation and lacking safeguards. The FCA’s Mills Review notes that “1 in 5 UK adults are already open to AI making decisions for them, with demand strongest where choices feel complex or high-stakes, particularly debt advice, pensions and investments”; and “around 26% trust general-purpose tools such as ChatGPT, Claude or Gemini for financial advice, despite limited awareness that formal routes to recourse will not apply”.^{[footnote 4]} 

These tools can provide outputs that resemble personalised advice without meeting standards on suitability, explainability, or accountability. There is no consumer protection, recourse, or redress where outcomes are poor, creating a material risk of harm across key financial decisions such as savings, investments, mortgages and retirement planning, as well as the potential for a loss of trust in AI. Consumers may also not know when they are engaging with regulated versus unregulated advice.  

This creates a clear asymmetry: regulated firms face strict obligations and liability, while unregulated AI providers can scale rapidly without equivalent safeguards, undermining both consumer protection and the intent of ongoing reforms such as Targeted Support. This dynamic is already beginning to materialise, and without intervention, risks shifting trust, innovation and consumer engagement away from regulated institutions into unregulated channels.           

If deployed within an appropriate regulatory framework, AI-driven advice has the potential to fundamentally reshape advice provision, enabling low-cost, personalised, and always-on support at population scale. This creates a clear strategic opportunity: to combine AI-driven triage, nudges, and guidance with human expertise for more complex needs, significantly expanding access without compromising quality. Realising this, however, requires consideration of the regulatory perimeter; clarifying the boundary between advice and guidance, and introducing proportionate guardrails for AI-enabled solutions to ensure accountability and consumer protection.  

While there are multiple viable approaches, there is strong consensus that a review of the regulatory perimeter must now be prioritised to introduce proportionate guardrails for AI-enabled services. We have intentionally not defined or prescribed what the next phase of this regulatory framework should look like, as we recognise there are multiple viable strategic and architectural approaches that could be considered by the government, informed by regulatory expertise and evidence. If done properly, this would create a level playing field across regulated and unregulated providers, unlock innovation, and allow trusted institutions to scale safe advice models. The prize is material: significantly improved consumer outcomes across financial resilience, better decision-making in savings and investments, and a more productive allocation of capital, delivering meaningful benefits for both consumers and the broader UK economy.      

This issue is not limited to financial services, with general-purpose models used to provide advice across a range of sectors, including healthcare, legal and tax.

3. AI Sovereignty and Resilience

To remain globally competitive and deliver pioneering innovation, UK banks currently have no viable option but to rely on a small number of global AI and cloud providers. While this adoption is necessary for immediate innovation, this dependence on a highly concentrated cluster of global tech providers (e.g. cloud and model hosts) introduces significant long-term concerns about operational resilience, data security, and concentration risk. By growing domestic AI research and solutions through partnerships, the sector can reduce these risks, ensure resilience, and spur local innovation without resorting to protectionism.      

Several large AI models and cloud service providers (often non-UK) are becoming systemically critical to financial firms, but they currently lie outside direct regulatory oversight. If one of these services fails or has a security breach, it could have widespread impact. Bringing such providers into the regulatory fold, for instance, via the Critical Third Party (CTP) designation powers, will ensure oversight keeps pace and important third-party services meet robust standards. In summary, the UK should strengthen its AI resilience by diversifying its technology base and ensuring critical external providers are subject to appropriate scrutiny, without adopting a protectionist stance that could hinder innovation.

4. Skills and Talent

In financial services, this requirement is particularly acute given the need to embed AI within regulated environments, requiring deep capability across model risk management, governance, and responsible deployment. 

The rapid advancement of AI demands a step-change in skills and talent across the UK workforce.  Financial services face additional challenges given strict regulatory, governance and risk management requirements that call for more specific and deeper AI skills. The sector needs people at all levels with AI-related skills, not just engineers and data scientists, but also leaders, risk managers, legal experts, and frontline employees who understand enough about AI to use it and govern it responsibly. Building this broad skills base will require significant upskilling and reskilling, as well as developing the future talent pipeline through education and training partnerships. 

A unified strategy will prioritise key skill areas, promote sharing of training resources, and keep the UK’s financial workforce globally competitive as AI evolves. It should also focus on making these opportunities accessible to people from all backgrounds and regions, so the benefits of an AI-skilled economy are felt widely across the country.

5. Agentic Payment Readiness

Agentic technologies present a significant opportunity to enhance how consumers engage with financial services, building on the UK’s high levels of digital adoption and enabling more seamless, personalised and responsive interactions in an increasingly digital-first market. Agentic payments provide a near-term, practical proxy for a broader class of emerging autonomous financial systems. 

Given the near-term focus of this Adoption Plan, engagement and recommendations have specifically considered agentic payments rather than broader, more complex agentic applications. While the long-term potential of these autonomous technologies is vast, the commercial reality brings specific challenges to the forefront. Firms have highlighted significant uncertainty around legal and regulatory accountability, including liability and consent, alongside heightened concerns about the potential for fraud within automated payment flows. 

Focusing on agentic payments provides a highly practical proxy for a wider set of emerging financial services use cases. It allows us to examine the friction points where multiple autonomous agents and third-parties operate across complex value chains - areas for which existing legal and regulatory frameworks do not yet provide sufficient clarity on responsibility and consumer protections. 

The immediate challenge is to clarify how existing legal and regulatory frameworks apply, or need to adapt, specifically for these near-term payment use cases. By resolving these foundational issues first, the UK can safely embrace cutting-edge, autonomous financial tools with confidence, creating strong consumer safeguards today that can scale to broader agentic applications in the future.

Recommended actions, prioritisation, and broader considerations

The following recommendations focus on practical steps to accelerate safe and effective AI adoption. They build on the existing UK framework and are intended to provide clarity and confidence for firms looking to scale AI. 

Within each theme, we set out a series of actions for government, regulators and industry, recognising that progress will require coordinated delivery across all parties. Implementation will need to remain iterative and responsive to ongoing technological developments, regulatory learnings and market feedback. As such, these actions are intended to provide a clear starting point for delivery, while allowing for refinement as the UK continues to develop its position as a global leader in responsible AI adoption in financial services.   

The recommendations and considerations below relate to three categories: 

• Immediate priorities: actions required to unlock near-term scaling (regulatory clarity, perimeter review) 
• Structural enablers: initiatives to support consistent adoption (assurance, skills, coordination mechanisms) 
• Strategic resilience: longer-term actions to manage systemic risk and competitiveness

Regulatory framework (recommendation 1)

1. Regulators should work together to ensure expectations of firms are clear and that services are accessible and navigable to support innovation

Feedback from industry highlighted that while regulators have established a range of valuable initiatives to support AI adoption,  they are not widely accessible to the majority of firms, and that it can generally be quite challenging to navigate the regulatory landscape across regulators, and the application of rules in the context of AI and agentic AI. We recommend that regulators consider what more they can do to provide accessible support to firms. 

Existing regulatory initiatives already represent positive steps that can be built on, including providing practical support to firms through the FCA AI Lab, Information Commissioner’s Office (ICO) engagement with industry on agentic AI, and collaboration through the Digital Regulation Cooperation Forum (DRCF). 

We recommend that regulators consider:  

• Confirming that the UK will maintain its principles-based, outcomes-focused regulatory approach for AI in financial services, while recognising that AI may raise novel questions that may warrant a targeted response. This will give confidence to firms that the UK’s regulatory approach is intended to support safe and responsible AI adoption within the existing framework, while allowing regulators to provide targeted clarification or adaption where needed as AI capabilities and use cases evolve.  
• Where collaboration could help to provide regulatory certainty to firms about how best to comply with expectations from different regulators (e.g. FCA, Prudential Regulation Authority (PRA), ICO, and CMA); and if there are ways to make regulatory initiatives and information more accessible, to help firms understand what’s most relevant for them. In particular, consideration of workshops with industry on areas of uncertainty, and improving firms’ access to clear joined-up regulatory support for AI Adoption. This should include clear regulatory information on how existing frameworks (e.g. Consumer Duty, PRA Supervisory Statement 1/23 on model risk management, Operational Resilience and Third-Party Risk Management, Senior Managers and Certification Regime (SM&CR) and relevant international standards or guidance) apply to common AI and agentic use cases. It may be helpful to also consider delivering support through mechanisms such as a Financial Services AI adoption Support Hub, with two potential functions: 
• Information Portal providing details of current initiatives (e.g. FCA AI lab updates, ICO agentic reports) 
• Effective access to supervisory and subject matter experts, for example on material novel use cases  

The government should continue to support cross-regulator collaboration, and invest in infrastructure such as compute and sovereign data infrastructure (see recommendations below). 

Priority: High

Regulatory perimeter (recommendation 2 - 3)

2. The FCA should undertake a comprehensive review of the consumer, competition and wider impacts of financial guidance and advice-like outputs generated by general purpose large language models (LLMs). Based on the findings, the FCA should work with the government to develop a clear policy and regulatory response

The FCA should review the consumer, competition and wider impacts from financial guidance and advice-like outputs generated by general purpose LLMs. This review should assess the benefits and risks of LLM-driven advice and guidance, including consumer outcomes and the potential for harm. It should also address industry concerns regarding an uneven playing field, and whether this risks distorting consumer behaviour and market competition. This should be achieved without unduly constraining innovation, or the customer benefits it can unlock in improving financial management. 

These findings should directly inform HMT’s consideration of the regulatory perimeter, including whether further consumer facing measures such as disclosures and education are needed, as they have been in response to fraud (building on recent publications such as the FCA’s InvestSmart article on using AI for investment research). 

Priority: High

3. Adopt a consistent consumer disclosure for AI-driven services.

The industry should explore developing a simple, consistent form of words to be used on a voluntary basis across the sector to help consumers identify regulated (from unregulated) AI enabled financial guidance and advice-like outputs.​ ​

Priority: High

Resilience (recommendations 4 - 6)

4. Accelerate implementation of the Critical Third-Party (CTP) regime, including assessment of Key AI/Cloud Providers

​​​Government (HMT) and regulators (Bank of England/PRA and FCA) to assess critical AI and cloud providers under the Critical Third Parties (CTP) regime, and where appropriate ​government should ​designate​ CTPs​, to ensure systemic risks are identified, monitored, and mitigated as adoption scales. 

Priority: High

5. Establish voluntary AI incident and “near-miss” sharing across the UK financial sector

​​​​The industry, potentially supported by the Cross Market Operational Resilience Group (CMORG) should c​​reate an industry-wide “AI Incident & Near-Miss” repository (similar to the approach on sharing cyber information) to share learnings across industry, which are unattributed to firms. This repository should capture unintended positive consequences, as well as incidents and near misses. ​The core focus is to establish a shared resource that fosters a culture of collective intelligence across the UK financial ecosystem. This initiative would complement not replace existing regulatory requirements on firms and, once designated, critical third parties (CTPs) to report certain incidents to the regulators and proactively notify them of matters of which they would reasonably expect notice.​​     ​​ 

Priority: Medium

6. ​​​Launch a voluntary, industry-led AI third-party assurance scheme for financial services, with a view to potentially working with regulators and government to standardise it in the future       

Building on existing AI assurance assessments already in use by firms, the industry​ – potentially supported by a central body such as CMORG -​ should explore​​ ​​developing an AI Third Party  Assurance Framework to enable consistent assessment of general-purpose or third-party AI models, systems, applications etc supporting independent audit and​ ​certification of AI third-party  providers that firms and regulators can rely on as a basis for assessing the relevant risks. For example, a qualified third-party assessor (or central body) would evaluate AI model providers against agreed standards (e.g., transparency, cybersecurity, reliability, data protection).​​ ​It will be critical that the framework reflects the diversity of AI and use cases, and is subject to regular review to ensure it remains effective and proportionate as technologies evolve. ​Regulators may then accept these assurance certificates as evidence of baseline due diligence by any firm using that model, reducing the need for firm-by-firm assessment questionnaires and documentation​, but each firm would remain responsible for appropriate risk management of implementing the model​. This would significantly reduce duplication of model risk assessments across firms, lowering costs, and accelerating safe adoption.      

By introducing a centralised AI assurance framework, the industry gains a standardised definition of a solid AI foundation, fostering alignment between firms and regulators. This approach streamlines compliance and eliminates duplicative work without compromising the rigorous, firm-specific requirements necessary for safe AI deployment. ​​This will help with overall cost reduction, less friction, and faster safe scaling of AI across the sector.  

The framework could operate as a voluntary, industry-approved scheme at first. However, in time it may be adopted by a central body, and regulators may choose to formally integrate its outputs into their supervisory approach – e.g., accepting a certified model as ​contributing to ​certain model risk management requirements. The framework should serve as a standardised audit protocol (similar to SOC2) to build institutional trust and ​reduce​ duplication across individual firms, who currently internally validate each model individually. 

The standards should be developed as soon as possible using the AI Assurance assessments firms will already have in use as a benchmark​.​ 

Priority: Medium

Skills and talent (recommendations 7 - 9) 

7. Encourage industry participation in the Financial Services Skills Compact and mobilise industry commitment 

HMT and the FSSC to drive industry participation in the Financial Services Skills Compact, securing voluntary commitments from leading firms to invest in AI training and capability-building. This would demonstrate leadership and commitment to investing in people, while encouraging collaboration to address the skills gaps required for safe AI adoption and deployment, and support growth. 

Priority: High

8. HMT to work in partnership with industry to build on the Financial Services Skills Commission’s research and recommendations to explore the development of a sector-wide financial services AI skills plan 

HMT to consider the Financial Services Skills Commission’s research, forthcoming recommendations and Skills Compact, and work in partnership with industry to explore the development of a sector wide AI skills plan to address skills gaps and support a highly skilled, AI fluent financial services workforce. This should ensure capability across boards, specialists and frontline teams. 

This should also include an approach to investment in education and training that equips individuals with relevant skills and expertise across lower and higher education to ensure the future talent pipeline. 

The plan should explicitly consider equitable access to reskilling and training, alongside regional delivery to support AI skills development across the entire current and potential workforce, as well as all socio-economic backgrounds and regions across the UK. 

The strategy should be kept under regular review and support long-term competitiveness.

Priority: High

9. Attract top global AI talent 

Reduce barriers to recruiting international AI specialists into UK financial services. HMT and the Home Office (in consultation with industry) should implement pragmatic adjustments to the existing visa framework to attract and retain global AI talent in UK financial services, building on the AI Opportunities Action Plan.  

This could include streamlining high-skilled visa processes, reducing costs and administrative burden, and considering whether dependent pathways could align with primary applicants. Specifically, reforms could address current frictions in existing routes (e.g. Global Talent visa), where feedback suggests processes are optimised for the primary applicant but remain more complex and slower for dependents, often requiring separate pathways or delayed access to employment. There should also be a focus on attracting international talent, ensuring partnerships and appropriate standards in key talent areas. Addressing these gaps and promoting the UK’s FS sector as a leading AI destination will be critical to improving competitiveness. 

Priority: Medium

​Agentic payments readiness (recommendation 10)

10. ​​​Leverage the upcoming HMT consultation to establish a ”trust framework” to support agentic payments protocol ​​ 

​​​​​​​​Agentic payments have significant potential to transform the global transaction landscape. While government, industry, and regulators are already laying the foundations for UK leadership through upgrades to payment infrastructure and digital money initiatives, achieving safe adoption at scale requires comprehensive standards.​​​ 

​​​​​​​​​​HMT’s upcoming consultation on modernising payment services regulation is a welcome step, but its outcome must address complex, systemic challenges. Rather than treating this solely as a regulatory update, the government, regulators and industry must collaborate standards built on three critical pillars:​​​ 

​​​​​​10.1. ​​​Legal & Liability Frameworks​: Defining clear legal constructs and dispute mechanisms to unambiguously assign accountability when autonomous agents transact.​​​

10​.2​​. ​​​Know Your Agent (KYA) Protocols​: Establishing standardised identity and verification frameworks specifically designed for AI and autonomous software agents.​​​ 

​​​​​​​10.3. ​​​Authentication & Governance​: Creating interoperable technical standards that ensure safe, frictionless, and trusted machine-to-machine authentication.​​​ 

​​​​​​​​​​​​Crucially, these standards should be informed by lessons from challenges within today’s payments ecosystem, including fraud. While the development of practical, pro‑innovation standards should first be led by industry, potentially with the support of a neutral body such as the Centre for Finance, Innovation and Technology (CFIT) that could convene the key stakeholders through its coalition model; there is also a valuable role for government and regulators in supporting this work and ensuring a clear regulatory framework — including, where appropriate, bringing forward legislative changes to provide clarity and enable safe adoption. With continued collaboration between government, regulators and industry, there is a strong opportunity to build momentum and reinforce the UK’s position as a global leader.​ 

Priority: High

Broader considerations for government: AI sovereignty and resilience

While our recommendations focus on specific challenges for financial services, there are cross-sector barriers that impact financial services, such as AI sovereignty and resilience. We are therefore proposing the following ideas that we suggest the government evaluate and consider when developing its wider AI strategy. These considerations focus on ensuring that across the economy the UK remains globally competitive, at the forefront of innovation, and how the government can reduce risks and ensure resilience.

1. Explore forging industry-academia partnerships to advance UK’s AI capability 

HMT and DSIT should explore fostering partnerships with industry and academia to accelerate domestic AI innovation and address sovereign risk, building on the investment set out in the Chancellor’s Mais Lecture 2026.

Financial services has significant potential to benefit from advances in AI capability, including through improved customer outcomes, enhanced fraud and financial crime detection, more effective risk management and increased productivity. Strengthening domestic AI capability through closer collaboration between industry and academia could accelerate the development of innovative UK-based solutions, helping the sector realise these benefits at greater scale while also supporting a more diverse and resilient AI ecosystem.

2. Build upon existing policy to communicate the UK’s Sovereign AI ambition and deliver an industry resilience strategy

The government is currently undertaking work to support the development of Sovereign AI capability in the UK. In developing that ambition, the government should work closely with industry to define and clearly communicate the strategic aims of this policy work. This is critical for driving the next phase of financial services adoption, managing systemic risk, and providing clarity on the UK’s long-term approach to non-domestic AI infrastructure. 

This long-term ambition must be framed not as a move toward nationalistic protectionism or mandated technology stacks, but as a strategic effort to guarantee market pluralism and systemic resilience. The government’s narrative should clarify that sovereign infrastructure is intended to expand - not restrict - commercial choice, providing a high-performance domestic alternative that mitigates the risk of US hyper-scaler concentration. 

To deliver on this, the government should focus on two action areas:

2.1 Clarify the strategic trajectory: HMG should articulate its approach regarding the mix of domestic capability and international reliance.

• To the extent that domestic infrastructure is developed, government should ensure that AI and energy policy are aligned so that new capacity supports the UK’s net zero objectives.
• To the extent that the UK relies on non-domestic infrastructure, the government must prioritise a plan to mitigate the resulting concentration risks, including by considering strategic AI trade partnerships with aligned international jurisdictions.
• Consideration should be given to ensuring portability between providers and access to compute capacity. These efforts should not restrict access to global technology, reduce competitiveness, or slow adoption, but rather ensure a baseline of systemic resilience.

2.2 Drive industry integration: Through further consultation with industry, the government should consider how to bring the FS sector and early-stage UK AI startups together using structured partnership frameworks and targeted commercial incentives. This will encourage financial services firms to safely use and scale less mature domestic AI providers.

As AI becomes more deeply embedded in financial services, reliance on a small number of overseas providers increases concentration and resilience risks. A clearer long-term strategy for sovereign AI would provide greater certainty for firms and support a more resilient and diversified AI ecosystem.

3. Strengthen the UK’s resilience to AI related risks across critical sectors

Reflecting on the critical coordination gaps for Frontier AI (e.g. Mythos), HMT and DSIT should consider working with the National Cyber Security Centre (NCSC) and the AI Security Institute (AISI) to establish a cross-sector AI Risk and Resilience Taskforce to eliminate silos and ensure sectors are proactively positioned to manage fast-moving, cross-sector vulnerabilities. 

A Taskforce could be structured around two core pillars: 

3.1. Cross-sector dependency mapping and horizon scanning: A suitable coordinating authority, such as the DSIT, could lead on efforts to co-ordinate across public authorities and industry, to build a common understanding of how AI-related risks and dependencies may transmit across critical sectors, including finance, energy, telecoms, cloud and digital infrastructure. This could, draw on relevant research and technical insights from bodies such as the AISI and NCSC while recognising existing channels for sharing such information. 

3.2. Cross-sector scenario testing and response planning: Government, regulators and industry could use that shared understanding to consider how AI-related scenarios which cause concurrent disruption across multiple critical sectors should be reflected in existing cross-sector resilience exercises, contingency planning and incident response arrangements. This should test whether existing arrangements remain sufficient, and help identify practical improvements to resilience planning without necessarily creating new regulatory expectations.

Financial service is highly interconnected with wider digital and operational infrastructure and can be affected by AI-related vulnerabilities that emerge beyond the sector itself. Improved cross-sector coordination would support earlier identification of emerging risks, faster dissemination of insights and a more consistent approach to strengthening resilience across the economy.

If more efficient, this could be incorporated into an already existing coordination group.

1. Department for Science, Innovation and Technology (DSIT) (2026)  AI adoption research. Published 13 February 2026 ↩

2. Bank of England (BOE) (2024) Artificial Intelligence in UK financial services – 2024. Published 21 November 2024. ↩

3. Financial Conduct Authority (FCA) (2026) Understanding the advice market: financial advice firms survey 2025. Published 23 April 2026. ↩

4. FCA (2026) The Mills Review : AI and the future of retail financial services. ↩