Bank of England Official Outlines AI Regulation Challenges

Bank of England Weighs Regulatory Approach for Autonomous AI in Finance

As artificial intelligence systems gain autonomy in financial markets, central banks are grappling with how to maintain stability without stifling innovation. Sarah Breeden, deputy governor for financial stability at the Bank of England, recently outlined the regulatory challenges posed by increasingly independent AI decision-making in a discussion with Wharton professor Itay Goldstein.

The conversation, part of Wharton's "Future of Finance" series, centered on what regulators call "agentic AI" — systems capable of making financial decisions with minimal human intervention. This shift represents a fundamental change from traditional algorithmic trading, where humans set parameters and AI executes within defined boundaries.

Why it matters

Financial regulators face a critical window to establish oversight frameworks before autonomous AI becomes deeply embedded in market infrastructure. The wrong approach could either create systemic vulnerabilities or push innovation offshore to less-regulated jurisdictions. Central banks must balance enabling technological advancement with preventing scenarios where interconnected AI systems amplify market shocks or engage in unintended coordination.

The Human Oversight Question

A central tension in AI regulation involves determining the appropriate level of human supervision. As AI systems become more sophisticated, the traditional model of human decision-makers using AI as a tool breaks down. Breeden and Goldstein explored where regulators should draw lines around autonomous operation versus mandatory human checkpoints.

The discussion also addressed whether existing regulatory tools remain adequate. Financial stability frameworks were designed for human-led institutions, not networks of AI agents that can execute thousands of decisions per second and potentially interact in unpredictable ways.

Measuring Regulatory Success

The conversation touched on how regulators will know if their AI oversight approaches are working. Traditional metrics like market volatility or institutional failures may not capture emerging risks from AI systems. Regulators need new indicators that can detect problems before they cascade through interconnected financial networks.

Breeden's participation signals that major central banks are actively developing positions on AI governance rather than waiting for problems to emerge. The Bank of England has been studying how AI could affect financial stability, including scenarios where multiple institutions deploy similar AI strategies that could amplify market movements.

Goldstein's research has examined how AI-powered trading affects market dynamics, including the potential for algorithmic collusion — where AI systems learn to coordinate behavior without explicit programming to do so. This work, conducted with Winston Wei Dou and Yan Ji, explores whether AI trading improves price efficiency or introduces new forms of market manipulation.

The discussion was produced by the Wharton Future of Finance Initiative and is available on major podcast platforms. Details were first reported by Wharton's Knowledge platform.