Bank of England: Current Rules Were Not Built for Agentic AI in Finance

In July 2026, the Bank of England announced a formal review of its regulatory frameworks to determine whether current rules cover the use of agentic AI in finance: in payment systems, securities trading, cybersecurity and operations.

Why current rules were not built for agentic AI

Bank of England Deputy Governor Sarah Breeden spoke at the European Central Bank's central banking forum and made a direct admission: existing regulatory frameworks were created without accounting for AI systems that can make decisions and act without a human directly involved at every step.

Agentic systems are a fundamentally different class of software. Unlike traditional algorithms or early language models, agents can break tasks down on their own, call external tools and initiate real-world actions: making payments, executing trades, adjusting risk parameters or responding to cyberattacks, all without explicit operator confirmation at each step. This autonomy creates a gap between the letter of current regulation and the way AI is actually used.

• Areas under review: payments, trading, cybersecurity, operational processes
• Breeden's key point: existing frameworks were not designed for autonomous AI agents
• Venue: ECB central banking forum, July 2026

Where the regulatory gap opens up

Most modern financial regulation rests on a basic principle: every significant financial decision must have an identifiable person who is accountable for it and able to explain it. Regulators require firms to document who made the decision, on what data, with what authority and within which supervisory framework.

Agentic systems push this logic to its limit. A single transaction or trade can involve dozens of autonomous steps carried out by several interacting agents: one gathers data, a second forms a hypothesis, a third executes the trade, a fourth manages risk. Where in this chain is the "responsible person"? Under current law, nowhere.

Speed adds another complication. Agentic systems in trading and cybersecurity make decisions in milliseconds. Full human oversight in real time is physically impossible, yet most existing accountability rules assume exactly that.

"Existing regulatory frameworks were not designed for AI agents capable of acting without direct human involvement," said

Sarah Breeden, Deputy Governor of the Bank of England.

What this means

The Bank of England is one of the most influential financial regulators in the world, and other jurisdictions look to its standards. Breeden's public admission at an ECB event is not an academic observation but an official signal: the regulator sees the problem and is starting to act.

For financial firms already deploying AI agents in trading systems, payment infrastructure and security teams, this means the period of quiet regulatory uncertainty is ending. What comes next is either targeted clarifications of existing rules or entirely new requirements for autonomous systems. It is better to prepare now.