As part of its “Financial Stability in Focus” work, the FPC has published a paper looking at how the rapid pace of development in firms’ use of AI can cause uncertainties and how these may translate into financial stability risks.
The report notes:
- how AI has the potential to save significant time and therefore boost productivity;
- how AI can enhance firms’ decision-making processes and so help make products and services better and more tailored to customer needs;
- that financial institutions are already using AI to automate and optimise internal processes and customer interactions – and that a likely future development area will be using it to inform core financial decisions such as credit and insurance underwriting, which could potentially shift the allocation of capital;
- specifically, AI can introduce risks in relation to models and data – which firms can mitigate by appropriate controls including “agentic AI”, but there is a risk that common weaknesses in popular models could cause a significant number of firms to misestimate certain risks and therefore misprice and misallocate credit;
- that a reliance on AI models for key decisions could lead to conduct-related risks if decisions or processes ended up being subject to litigation or redress
- use of AI to inform trading and investment decisions could increase market efficiency but also lead to collective decision making that would reduce stability if firms take increasingly correlated positions, which would lead to issues of availability and funding;
- reliance on a small number of key providers would lead to operational risks and by extension systemic risks;
- AI can both increase defences against cyberattacks while also increasing malicious actors’ ability to carry them out; and
- more widely, if AI challenges established business models in certain sectors, this could impact firms in those sectors being able to borrow.
The report looks more closely at:
- greater use of AI in bank and insurer core decision making
- greater use of AI in financial markets
- operational risks in relation to AI service providers and
- changing external cyber threat environment.