The financial sector's increasing reliance on generative artificial intelligence (AI) is generating new concerns about the stability of financial systems, according to Sarah Breeden, Deputy Governor of the Bank of England. Speaking at a conference in Hong Kong, Breeden highlighted the rapid growth in AI utilisation by lenders, which is prompting the central bank to consider incorporating AI-related risks into its annual stress tests.
Breeden emphasised the swift expansion of AI's scope and influence, noting that the technology is now being utilised by 75% of financial companies, a marked increase from 53% two years ago, according to a recent Bank of England survey. This accelerated adoption of AI is particularly notable in applications involving automated decision-making, with more than 50% of use cases integrating some level of automation.
The generative AI systems, capable of producing text, code, and videos almost instantaneously, are viewed by the Bank of England as a potential source of market manipulation and volatility. These concerns are particularly pertinent when AI is employed in trading, where it could exacerbate market fluctuations, especially during periods of stress.
In response to these potential risks, the Bank of England is formulating an "AI consortium" in collaboration with private sector experts to thoroughly examine these risks. The central bank is considering applying insights from this consortium to its stress tests, which aim to evaluate the resilience of UK banks under various crisis scenarios. This initiative seeks to understand better how AI models might interact during trading activities, whether utilised by banks or non-banks.
Breeden cautioned against the risks associated with "crowded trades" funded through leverage, noting that market shocks could lead to significant losses in these strategies. This, in turn, could trigger forced selling and unfavourable price movements, amplifying market stress via feedback loops.
These concerns align with a warning issued by the International Monetary Fund (IMF) in its recent financial stability report, which drew attention to the potential for AI to accelerate market fluctuations and increase volatility in times of market stress.
Sarah Breeden, who has been in her role since November last year, stressed the importance of accountability within financial firms, suggesting that regulatory rules might need adjustments to ensure senior bankers remain responsible for AI-driven decision-making. She highlighted the necessity for managers to thoroughly understand and manage AI models as these evolve autonomously.
While many AI applications in financial services present relatively low risks from a stability perspective, Breeden highlighted emerging significant use cases, such as credit risk assessment and algorithmic trading, which warrant close attention. The Bank of England survey revealed that 41% of companies are utilising AI to optimise internal processes, over a quarter for customer support, and at least a third to combat fraud. Notably, 16% are using AI for credit risk assessment, with 19% planning to adopt it within the next three years.
The survey also indicated that 11% of financial firms are engaging in algorithmic trading with AI, with an additional 9% intending to do so within the next three years. Breeden pointed out that roughly half of AI applications in financial companies involve semi-autonomous decision-making with human oversight, while the rest are fully automated, presenting significant challenges for management, governance, and regulation.
These developments showcase the dual nature of AI's impact on the financial sector: its potential to enhance efficiency and profitability, alongside the risk of introducing complex challenges that could affect market stability and resilience.
Source: Noah Wire Services