The FCA has published a summary of the AI sprint it held in January, which had 115 participants. 4 common themes came out of the event:
- that firms need clarity from the FCA on how regulatory frameworks apply to AI;
- the importance of trust – without trust, the full benefits of AI in financial services can’t be recognised;
- collaboration is key – between regulators, governments, firms, academics, model developers and end users; and
- a safe testing environment is very helpful.
Over the next 5 years, firms see the possibilities for development of virtual AI assistants and AI financial advisers and how agentic AI could help to personalise and scale up consumers’ interactions with financial services, and how AI can increase efficiency and effectiveness. They saw use cases of AI financial advisers, internal process automation, compliance tools, emotion AI and personalisation and agents.
Participants discussed how to ensure safe and responsible AI adoption – and focused on:
- success measures being clear – and that AI use should be held to standards at least as high as humans
- clear model, data, cloud and technology foundations being key to AI innovation
- the need to upskill staff on AI and internal governance and
- collaborating to develop common data standards and definitions
In terms of the regulatory framework, firms discussed:
- robust processes within firms: entailing effective internal governance and clear accountability for AI use cases, the helpfulness of assurance processes and recognised standards, and how regulatory requirements around outsourcing and critical third parties mitigates risks;
- good outcomes for consumers: here the key is transparency and whether more should be done to ensure fair outcomes given the inherent potential risks of bias on vulnerable consumers, and also whether AI could deliver further benefits to consumers; and
- effective competitive markets: generally there is currently uncertainty on the impact of AI on competition in markets. There is a concentration risk if a small number of firms develop models but AI can also help reduce barriers to entry.
The FCA will focus on:
- providing a safe space to innovate through its “Supercharged Sandbox”;
- addressing areas of uncertainty;
- fully participating in and effectively influencing international standard-setting bodies and collaborating on a bilateral basis with other regulators; and
- providing clear communication on its approach to AI.