The Bank of England Just Called for AI Circuit Breakers. Every Board Needs to Understand Why.

Bank of England Deputy Governor Sarah Breeden, speaking at the ECB Forum on Central Banking in Sintra on June 30, 2026, called for circuit breakers and kill switches on agentic AI systems operating in financial markets. She identified three governance requirements: automated circuit breakers that can pause or halt AI agent activity when risk thresholds are breached, human-in-the-loop requirements for high-stakes financial decisions, and systemic risk monitoring frameworks that can detect when multiple AI agents are behaving in correlated ways that could amplify market instability.

An AI circuit breaker is an automated governance mechanism that pauses or terminates AI agent activity when predefined risk thresholds are exceeded. In financial markets, this could mean halting an AI trading agent when its position size exceeds a risk limit, pausing an AI credit assessment system when its outputs deviate significantly from historical patterns, or stopping an AI customer service agent when it begins generating responses outside its authorised scope. Circuit breakers are designed to prevent AI systems from taking actions that compound errors or amplify systemic risk before human oversight can intervene.

Agentic AI systems are qualitatively different from earlier AI tools because they can take sequences of actions autonomously, interact with other systems, and pursue objectives over extended time periods without continuous human supervision. The Bank of England's concern is that agentic AI in financial markets creates new systemic risk vectors that traditional regulatory frameworks were not designed to address. Specifically, multiple AI agents from different institutions could behave in correlated ways during market stress, amplifying volatility in ways that no single institution's risk management would detect. The circuit breaker and kill switch framework is designed to address this systemic dimension.

The Bank of England's governance framework for agentic AI in financial markets is a leading indicator for regulatory expectations across all sectors. Financial services regulators consistently set the governance standards that other regulators follow, particularly for systemic risk, operational resilience, and third-party risk management. Enterprise boards in technology, healthcare, professional services, and critical infrastructure should treat the Bank of England's circuit breaker framework as a preview of the governance requirements they will face within the next 12 to 24 months. Boards that build this infrastructure proactively will be better positioned when sector-specific regulation arrives.

A circuit breaker is a temporary pause mechanism that halts AI agent activity when a risk threshold is breached, with the expectation that activity can resume after human review and clearance. A kill switch is a permanent or extended termination mechanism that stops AI agent activity entirely, typically used when a system is behaving in ways that cannot be safely corrected through a temporary pause. In practice, most governance frameworks use circuit breakers as the primary control, with kill switches reserved for situations where the AI system's behaviour represents an unacceptable ongoing risk. The Bank of England's framework calls for both, with clear criteria for when each should be triggered.

Enterprise risk committees should treat agentic AI as a new risk category that requires dedicated governance infrastructure, not as a subset of existing IT or operational risk. The practical steps are: first, map all agentic AI systems currently in use or under development, including third-party AI agents accessed through APIs or platforms; second, define the risk thresholds that would trigger circuit breaker activation for each system; third, establish human-in-the-loop requirements for the categories of decision where AI agent errors would have material commercial or regulatory consequences; fourth, implement monitoring that can detect correlated behaviour across multiple AI systems; and fifth, document the governance framework in a format that satisfies regulatory disclosure requirements.

The UK regulatory landscape for agentic AI governance is evolving rapidly. The Financial Conduct Authority and Prudential Regulation Authority have both signalled that existing operational resilience frameworks apply to AI systems, including agentic AI. The Bank of England's June 2026 speech represents a significant escalation in the specificity of regulatory expectations. While formal rules have not yet been published, regulated financial institutions should treat the circuit breaker and kill switch framework as a strong signal of forthcoming supervisory expectations. The EU's DORA regulation, which came into force in January 2025, provides a parallel framework that UK regulators are likely to reference in developing domestic requirements.

The Bank of England's circuit breaker framework and DORA's ICT risk management requirements are complementary governance frameworks that address overlapping risks from different regulatory perspectives. DORA's Articles 10, 11, 13, and 26 establish requirements for ICT risk detection, response, learning, and threat-led penetration testing that apply directly to AI systems used by financial institutions operating in the EU. The Bank of England's framework adds the systemic risk dimension that DORA does not fully address: the risk that multiple AI agents across different institutions behave in correlated ways that amplify market instability. UK-regulated financial institutions with EU operations need to satisfy both frameworks simultaneously.