The Bank of England is currently reviewing whether existing financial regulations are adequate to govern the use of agentic AI systems. These autonomous AI agents operate with limited human oversight across various financial sectors, including payments, trading, cybersecurity, and operational workflows.
Challenges of Agentic AI in Finance
Deputy Governor Sarah Breeden highlighted that current regulatory frameworks were not originally designed to handle AI systems capable of independent decision-making and task execution without direct human instruction. She noted that expecting human oversight for every action of these AI agents is impractical, especially as their use expands.
Agentic AI differs from traditional automated tools by pursuing objectives and making decisions with minimal human intervention. Such systems are increasingly applied in product recommendations, process automation, and trading-related activities.
Cybersecurity and Financial Stability Risks
Breeden emphasised cyber resilience as a major concern. Advances in AI have significantly enhanced capabilities to identify and exploit cyber vulnerabilities rapidly and at scale. While AI tools can bolster cyber defences when used by security teams, the same technology could amplify risks if leveraged by malicious actors, potentially threatening financial stability.
Open-source AI models are closing the gap with advanced closed models, with only a few months difference in capability, limiting regulatory comfort despite some restrictions on advanced AI releases.
Regulatory Responses and Safeguards
The Bank of England is exploring stronger recovery requirements for core banking systems, including options for one bank to temporarily support another during outages. Other considerations include failover systems and rapid rebuilding capabilities for compromised infrastructure.
Regulators are also investigating guardrails such as circuit breakers and kill switches to halt trading if AI malfunctions cause severe market disruptions. This is crucial as autonomous systems could amplify market volatility by reacting similarly to shared signals, potentially diverging from intended goals or public policy objectives.
In June, the Financial Stability Board (FSB) released a consultation proposing 12 sound practices for responsible AI adoption in financial institutions. These focus on governance, risk management, and cyber and third-party risks related to AI, though they are not binding international standards.
Outlook
The Bank of England’s review aims to ensure financial firms remain resilient amid increasing use of agentic AI. This includes firm-level controls and market-wide safeguards to manage emerging risks.
For businesses seeking practical AI adoption strategies aligned with evolving regulations, JASON AI offers comprehensive solutions. Learn more at https://jasonjuul.com.
Disclaimer: This article provides an overview of regulatory developments and does not constitute legal or financial advice. Organisations should consult relevant experts when implementing AI governance frameworks.