The head of the Bank of England has warned that a new generation of artificial intelligence models poses a growing threat to the stability of the global financial system, according to comments reported by CNBC.
The intervention marks one of the starkest assessments yet from a leading central banker on the risks created by the rapid adoption of advanced AI across banking, trading and asset management. While regulators have spent the past few years cautiously welcoming AI as a source of efficiency and productivity, the tone from supervisors has shifted as the technology has moved from back-office automation into decision-making roles at the heart of markets.
Why regulators are worried
The central concern is not that any single AI system will fail, but that many firms may end up relying on very similar models. If banks, hedge funds and trading platforms all use a small number of powerful models built by a handful of technology providers, their strategies could converge. In calm markets, that concentration may go unnoticed. In a shock, it raises the prospect of institutions rushing for the same exit at the same time, amplifying price swings rather than absorbing them.
Related worries include the opacity of the most advanced systems, which can make it difficult for risk managers and supervisors to explain why a model behaved as it did; the potential for AI-driven trading to react faster than human oversight can intervene; and the operational dependence of the financial sector on a small number of AI and cloud computing suppliers, which creates single points of failure outside the regulatory perimeter.
There is also the question of misuse. Cheaper, more capable generative models make sophisticated fraud, impersonation and disinformation easier to produce at scale â a threat not only to individual customers but potentially to confidence in institutions themselves, at a time when deposits can move at the speed of a smartphone.
A familiar pattern of risk
Financial history offers uncomfortable parallels. Innovations that appeared to reduce risk at the level of the individual firm â from portfolio insurance to structured credit â have repeatedly increased fragility for the system as a whole by encouraging herding and false confidence in models. Central bankers have increasingly framed AI in similar terms: a technology with real benefits, but one whose systemic consequences are poorly understood and difficult to test in advance.
The Bank of England’s Financial Policy Committee has been examining AI-related vulnerabilities as part of its regular assessment of threats to the UK financial system, and international bodies including the Financial Stability Board and the Bank for International Settlements have flagged comparable concerns.
What happens next
Regulators face a difficult balance. Heavy-handed restrictions could push AI activity into less supervised corners of the market or leave domestic firms at a competitive disadvantage. Doing too little risks allowing dependencies to build quietly until they are too entrenched to unwind safely.
The likely direction of travel is greater disclosure from financial firms about where and how AI is being used, tougher expectations on model governance and human accountability, and stress testing designed to capture scenarios in which automated systems behave in correlated and unexpected ways. Supervisors are also pressing for clearer oversight of the third-party technology providers on which the industry increasingly depends.
For markets, the message from Threadneedle Street is less a prediction of imminent crisis than a warning against complacency: the tools reshaping finance are advancing faster than the frameworks built to contain them. Read More
Leave a Reply