In January 2025, we hosted a two-day artificial intelligence (AI) Sprint. Find out how we are using feedback from this event and how we want to continue working together.
The AI Sprint is one component of our AI Lab[1], which continues to support firms’ implementation of AI use cases and provide a pathway to engage with firms and wider stakeholders about AI-related insights, discussions and case studies.
About the AI Sprint
The AI Sprint discussed the opportunities and challenges of AI in financial services. 115 participants took part from across industry, academia, regulators, technology providers and consumer representatives.
In teams, they discussed how AI may develop in financial services over the next 5 years, and the FCA’s role enabling firms to embrace the benefits of AI while also managing the risks.
We summarise the feedback received alongside the responses to the AI Input Zone. We are reviewing this feedback to consider how we can further support firms looking to adopt AI in a safe and responsible way and encourage beneficial innovation and growth.
Key insights
During the AI Sprint, 4 common themes came from participants’ discussions and suggestions:
Phases explored
This Sprint focused on 2 main phases for participants to explore:
- The next 5 years of AI in financial services.
- The current financial services regulatory regime.
Phase 1: The next 5 years
Teams first considered how AI might develop in financial services and how current use cases may change and emerge. Findings included:
- Increasing personalisation for consumers: participants discussed virtual AI assistants and AI financial advisers. They explored how advancements in agentic AI (i.e. autonomous models) could help personalise and scale-up consumers’ interactions with financial services, leading to better outcomes.
- Increasing automation for firms: participants highlighted opportunities to increase the efficiency and effectiveness of internal processes. This included using AI to automate regulatory compliance, enhance fraud detection and help customer support agents.
Teams then discussed what conditions might be needed over the next 5 years to enable safe and responsible AI adoption. These included:
- Being clear on success measures: participants discussed adopting a measurable way of assessing whether AI should be used in certain circumstances. For example, AI use should ‘at least be held to as high standards as a human process’. This was linked to potential preconceptions of mistrust in AI systems.
- Solid foundations: participants discussed how essential clear model, data, cloud and technology foundations are for AI innovation. Upskilling staff on AI and effective internal governance were also highlighted as requirements for success.
- Collaborating: to develop common data standards and definitions could enhance interoperability, which would create more opportunities for innovation. Transparency and data sharing were highlighted as crucial to achieving this.
Phase 2: The financial services regulatory regime
Teams then looked at topics related to the current regulatory framework for financial services, which were split into 3 overarching themes.
Theme: Robust processes within firms
Governance & accountability
Governance & accountability: participants discussed the importance of effective internal governance and clear accountability for AI use cases within firms, including under the Senior Managers & Certification Regime (SM&CR). They noted challenges for firms in assigning risk ownership, defining controls and getting the appropriate information flows.
Assurance
Assurance: teams highlighted how assurance processes can help make models more effective and increase trust. It was noted that international standards for AI models are being developed and adopted, eg ISO 42001.
Outsourcing / critical third parties (CTP) requirements
Outsourcing / critical third parties (CTP) requirements: participants also assessed the current regulatory approach for third-party AI systems, considering how the existing approach mitigated risks and harnessed opportunities.
Theme: Good outcomes for consumers
Explainability and transparency
Explainability and transparency: participants suggested that consumers should be ‘in the loop’ by having clear explanations about how AI models are used, how they reach their decisions and have clear routes to escalate to a human or for redress if required.
Fairness (bias and vulnerability)
Fairness (bias and vulnerability): participants discussed whether there is more that should be done to ensure fair outcomes for consumers in the context of AI noting the inherent potential risk of bias and impacts on vulnerable consumers.
Consumer benefits
Consumer benefits: participants also considered if more could be done to harness the benefits of AI for consumers. They discussed how some of the emerging use cases (see phase 1) could improve customer experience - for example, an AI agent that helps pay down personal loans and save. This would enhance trust as well as create benefits for markets and businesses.
Theme: Effective competitive markets
Market risks
Market risks: participants reflected on the risks AI could create or enhance for financial markets, such as algorithmic collusion or lack of transparency. Many of these are already covered by existing principles and outcomes-based regulation.
Competition within markets
Competition within markets: participants highlighted uncertainties around the impact of AI on competition in markets. There is a concentration risk if infrastructure or models are developed by a small number of firms. However, AI can help reduce barriers to entry and support a thriving competitive sector.
UK growth and competitiveness
UK growth and competitiveness: participants also considered opportunities to harness AI to support the growth and competitiveness of UK financial services. They noted the UK already has the advantages of a robust existing regulatory framework, highly skilled talent, advanced research institutions, a sophisticated capital ecosystem and an appetite for AI adoption.
Next steps
We are committed to building on the momentum from the AI Sprint to support the safe and responsible adoption of AI across UK financial services.
The feedback and insights received during the AI Sprint – as well as on the AI Spotlight and AI Input Zone – have provided valuable information. They are being used to shape our future work on AI in line with our statutory objectives.
The following are key issues frequently raised by stakeholders that we are focusing on:
- The need for a safe space to innovate. In the coming months, we’ll be launching the Supercharged Sandbox, offering innovators greater computing power, infrastructure, datasets, and mentorship to support the testing and validation of AI solutions. The programme is being shaped by insights from the Sprint and identified focus areas to address key adoption challenges. The goal is to equip innovators with the necessary tools, expertise, and regulatory engagement to drive AI adoption in financial services.
- Considering whether there are areas of uncertainty where regulation could be restricting safe and responsible AI adoption.
- For example, one identified area of uncertainty was around data protection and privacy. With the ICO, we have since written to trade associations[3] to obtain more information about these persistent challenges and how we can better support industry in tackling them.
- International engagement: Ensuring we effectively influence the international standard-setting bodies work (e.g. IOSCO, FSB, GFIN) on AI to ensure we can support safe and responsible adoption in the UK.
- Bilateral engagements with other regulators to explore cross-cutting considerations as well as collaboration opportunities on specific themes.
- Clear communication about our approach to AI and continued engagement with stakeholders through our AI Lab[4]. The AI Spotlight will run on an ongoing basis, where new themes will be added based on the use cases discussed in the Sprint.
