Bank of England signals regulatory overhaul for autonomous AI

The rise of agentic AI marks a departure from traditional automated tools, as these systems pursue objectives independently rather than following static instructions. According to a 2026 report from the Cambridge Centre for Alternative Finance, 52% of financial firms are already deploying agentic AI, primarily for internal operational tasks. Breeden noted that while current adoption in trading remains focused on lower-risk processes, the technology's ability to identify and exploit cyber vulnerabilities represents a significant step change in risk.

Cyber resilience has become a primary stability concern, as malicious actors could leverage similar high-speed tools to trigger correlated failures across shared digital infrastructure. The International Monetary Fund has echoed these alarms, suggesting that attacks on shared cloud services or payment networks could lead to sector-wide disruption. To mitigate these threats, the Bank of England is evaluating new recovery requirements, including potential mandates for banks to assume basic functions for compromised peers during outages.

Regulators are also exploring technical safeguards such as circuit breakers and kill switches to curb volatility caused by AI models reacting in unison to market signals. While the Bank of England previously maintained that existing rules were sufficient, Breeden acknowledged that recent advancements have exposed critical gaps. The Financial Stability Board has similarly called for more robust governance, emphasizing that the shift toward autonomous finance requires moving beyond individual firm oversight to address risks inherent in the broader financial ecosystem.