June 30, 2026, (Inside AI) — The Bank of England’s deputy governor for financial stability, Sarah Breeden, warned on Tuesday that existing regulatory frameworks may be inadequate for overseeing autonomous AI agents in finance. She spoke at the European Central Bank Forum on central banking in Portugal.
Breeden argued that current rules were not designed for systems that act independently. She stressed that relying on human oversight for every agent action is unrealistic.
"Our frameworks were not built to contemplate autonomous agents, and relying on a human in the loop for all agent actions is unlikely to be realistic. More sophisticated governance and accountability frameworks may be needed," Breeden said.
Her remarks come amid heightened concern over AI risks. Since Anthropic released its Mythos model, analysts have flagged serious cybersecurity threats to banking. The Financial Stability Board earlier in June called for tighter safeguards against AI agents, citing distinct challenges to human oversight.
Breeden’s intervention signals a shift in regulatory thinking. Traditional models assume a human decision-maker at the end of every process. Agentic AI breaks that assumption by executing complex tasks without step-by-step human approval.
This creates a gap in accountability. If an AI agent makes a faulty trade or exposes customer data, who is liable? The developer, the deploying bank, or the agent itself? Current laws provide no clear answer.
The Bank of England is not alone in grappling with these questions. The European Central Bank has also explored agentic AI risks in its supervisory work. But no major jurisdiction has yet proposed concrete rules for autonomous financial agents.
Breeden’s call for more sophisticated frameworks suggests a need for technical standards, audit trails, and possibly mandatory kill switches. She did not outline specific measures but implied that governance must evolve alongside the technology.
Some industry voices push back, arguing that over-regulation could stifle innovation. They contend that existing risk management principles can be adapted. Yet Breeden’s remarks indicate that central bankers see a qualitative difference in agentic AI.
The challenge is global. AI agents do not respect borders, and a regulatory patchwork could create loopholes. Coordination through bodies like the FSB will be critical.
Breeden’s speech did not propose immediate rule changes but served as a clear signal. As AI agents become more capable, the window for proactive regulation is narrowing.
