Bank Of England: Artificial Intelligence Consortium Minutes – May 2025
The Artificial Intelligence Consortium (AIC) aims to provide a platform for public-private engagement to further dialogue on the capabilities, development, deployment, use, and potential risks of artificial intelligence (AI) in UK financial services. As stated in the AIC’s Terms of Reference, the views expressed by the members in these minutes and any subsequent outputs do not reflect the views of their institutions, the Bank or FCA. The activities, discussions, and outputs of the members should not be taken as an indication of future policy by the Bank or FCA.
Item 1: Introduction to the AI Consortium
Co-Chairs Sarah Breeden and Sarah Pritchard welcomed the members and observers to the inaugural meeting of the AIC.
Sarah Breeden highlighted members’ contributions would be under Chatham House Rule and reiterated members are acting in a personal capacity, rather than representing their organisations. Sarah emphasised the AIC is a forum designed to be member-led and recognised the importance of public-private engagement in ensuring AI is adopted safely and responsibly in financial services.
Sarah Breeden provided an overview of the Bank’s approach to AI, underscoring its importance to the Bank’s and FCA’s objectives, including maintaining financial stability and enhancing market integrity. Sarah highlighted the importance of a macroprudential perspective, overseen by the Bank through the Financial Policy Committee, which had outlined its view on the financial stability implications of AI in a recently published Financial Stability in Focus report. She drew attention to the joint work of the Bank and the FCA who established the AI Public-Private Forum (AIPPF) in 2021 designed to share information, deepen collective understanding of AI and explore how to support its safe adoption. The AIC will build on the efforts of the AIPPF.
Sarah Pritchard noted the fast pace of AI developments and reminded participants of the FCA’s commitment to an agile regulatory approach. Sarah reiterated the FCA’s outcomes-focused approach and provided an overview of various FCA AI initiatives, including the AI LabOpens in a new window . The AI Lab includes different components – such as AI Live TestingOpens in a new window and the Supercharged SandboxOpens in a new window – that are open to help firms test and experiment with AI use cases so that they have the confidence to adopt AI safely and responsibly.
During breakout sessions, members discussed AI topics of interest, associated risks and opportunities, and explored ways in which members could collaborate to produce useful outputs. Suggestions to collaborate included sharing tangible AI use cases and sharing knowledge through examples of good and bad practice to facilitate successful AI adoption. Members also discussed producing resources for non-technical practitioners to understand AI’s capabilities and producing reports on how to manage the risks and opportunities identified by the AIC. Members discussed a range of challenges and risks that the AIC could explore. Key concerns included the growing reliance on third-party providers, raising issues around concentration and accountability. They also noted that widespread use of similar AI models could amplify systemic vulnerabilities, particularly under market stress. Risks of contagion were highlighted, where flaws in one model might spread through the system to another model. There was also unease about generative AI introducing misleading information into financial markets, potentially distorting market signals and movements. Concerns around fairness, such as biased credit scoring, and the increasing threat of AI-driven fraud and cyberattacks were also raised. To address risks posed by AI, members highlighted several potential strategies and mitigations. These included identifying good practices for managing model-related risks, alongside approaches for monitoring AI usage, risks, and benefits – one member suggested exploring whether formats such as regulatory reporting could support supervisory oversight. Governance was another key theme, with some pointing to the role of international standards in supporting the safe adoption of AI. Some members highlighted the varying levels of technical expertise across the sector and thought education and awareness particularly for consumers and smaller firms were essential. Finally, members noted tangible benefits of AI in financial services, such as fraud detection, scaling personalised services, and generating synthetic market data to support model development. Members shared insights on the benefits and opportunities likely to dominate in the coming year. These included improving operational efficiency, enhancing customer experience, and driving organisational transformation through task automation. They also noted the potential for more advanced applications, such as agentic AI, to emerge, alongside the increased use of AI in fraud prevention. Sarah Pritchard asked the AIC members whether there was anything the Bank and FCA should uniquely focus on. Members noted both regulators have taken a pragmatic yet flexible approach to regulation so far. Multiple members highlighted the need to coordinate across other regulators, jurisdictions and sectors, noting it was useful to have other UK regulators in attendance at the AIC as observers. One member suggested introducing AI at an earlier stage of curriculums in education. The Co-Chairs and members agreed the next meeting would be held virtually in July and would focus on preparation for the AIC workshops, including by determining workshop topics and outputs. Members agreed under the AIC’s Terms of Reference to form small working groups to conduct in-depth and/or technical examinations of specific subjects and report back to the AIC. The subsequent AIC meeting was proposed for September. The Co-Chairs thanked the members for their participation. Breeden, Sarah – Bank of England Pritchard, Sarah – Financial Conduct Authority Ahmed, Ratul – Commerzbank AG Beliossi, Giovanni – Axyon AI SRL Bhatti, Tanveer – Revolut Buchanan, Bonnie Gai – University of Surrey Daley, Sue – techUK Dunmur, Alan – Allica Bank Hughes, Clara – Pension Insurance Corp Jefferson, Michael – Amazon Web Services Jones, Matthew – Nationwide Kazantsev, Gary – Bloomberg LP Kazim, Emre – Holistic AI Li, Feng – Bayes Business School Magazzeni, Daniele – JPMorgan Pearce, Christopher – esure Group Pearce, Luke – Santander Prince, Emily – LSEG Rees, Harriet – Starling Bank Limited Rosenshine, Kate – Microsoft Szpruch, Lukasz – The Alan Turing Institute Valane, Jeffrey – HSBC Wade, David – Goldman Sachs Wintle, Anna – Visa Xu, Justin – Millennium Global Investments Croxson, Karen – CMA Fairburn, James – HMT Seiler, Chia – OfcomItem 2: Potential topics for workshops and outputs of the AI Consortium
Challenges and risks
Potential strategies and mitigations
Benefits and opportunities
Summary of discussions
Item 3: Next steps
Attendees
Co-chairs & Moderators
Members
Observers