The Bank of England has set its sights on reviewing the regulation of AI agents: the stakes are rising
The Bank of England has officially announced its intention to radically overhaul the regulatory framework for artificial intelligence in the financial sector. Sarah Breeden, Deputy Governor of the regulator, speaking at the European Central Bank forum in Sintra, noted that current technology-neutral rules are not designed for systems capable of acting autonomously, without human intervention. She identified cyberattacks, payment system failures, and synchronous market errors as key threats from the widespread adoption of AI agents.
AI Acceleration: A Dual Challenge for Regulators
"The obvious acceleration in AI capabilities, even compared to the situation six months ago, creates a dual challenge: not only ensuring responsible deployment and managing risks, but also recognizing that AI must transform the way central banks operate. We need to seriously consider both the first and second aspects," Breeden emphasized.
Among the potential measures under consideration by the regulator are tightening requirements for restoring critical systems, creating fully isolated backup capacities, and the ability to rapidly rebuild compromised infrastructure. For financial markets, the Bank of England is studying mechanisms similar to trading circuit breakers to prevent large-scale disruptions caused by erroneous AI models. Of particular concern is the "herding behavior" of agents that, reacting to identical triggers, can amplify volatility.
International Coordination
On May 15, the Bank of England, together with the Financial Conduct Authority and HM Treasury, published a statement on the cyber risks of advanced AI models for the financial sector. On June 22, a similar assessment was provided by the Five Eyes, warning that the window between vulnerability discovery and exploitation is shrinking to months. On June 10, the Financial Stability Board presented a consultation document with 12 practices for responsible AI deployment, proposing a framework for corporate governance and risk control.
Deployment Reality: 81% Already in Play
According to data from the Cambridge Centre for Alternative Finance, 81% of surveyed industry participants have already implemented AI at some level, and 52% are in the pilot or more mature deployment stage of agentic AI. Currently, the primary application is focused on internal processes: automation, data visualization, and software development. Front-office deployment remains limited.
In a response to a parliamentary committee on April 1, the Bank of England acknowledged that generative and agentic AI do not yet pose a systemic risk but emphasized that the technology-neutral approach will be revised. In May 2026, the International Monetary Fund called for considering AI cyber risks as a matter of financial stability, not just IT security.
My analysis: The Bank of England's actions are not merely a bureaucratic response but a signal that the era of "regulatory lag" for AI is ending. The market should prepare for the emergence of stringent requirements for AI system architecture, particularly regarding the isolation of critical functions and "kill switch" mechanisms. For the crypto sector, where autonomous agents are already actively used in DeFi and trading, this may mean inevitable adaptation to new standards, which in the long term will increase resilience but in the short term will create operational costs.