Key Takeaways
- The FCA will adapt its regulatory framework to supervise AI-enabled financial systems and AI agents
- 62% of financial services firms have deployed AI agents, with 93% granting them autonomy
- Regulators worry that risks shift from single-firm harm to system-wide instability as AI autonomy grows
The UK's Financial Conduct Authority has laid out its playbook for regulating AI in banking, and the focus is clear: autonomy. In a June 2026 report, the FCA's executive director Sheldon Mills warned that as AI agents move from recommending actions to executing them, regulatory risk shifts from individual firms to the financial system itself. Banks are already racing ahead with deployments. Regulators are trying to keep pace.
The Mills Review identifies four ways AI will transform financial services: deeper integration into business operations, customer journeys led by AI agents rather than humans, shifting competitive dynamics among firms, and heightened fraud threats. That last point matters because AI agents acting on behalf of customers create attack surfaces that did not exist when humans handled every transaction.
How widespread is AI agent adoption in finance?
Widespread, and accelerating. A June 2026 report from the Cloud Security Alliance found that 62% of financial services firms have deployed AI agents. Of those, 93% have granted their agents some level of autonomy. That is not a pilot program statistic. It is mainstream adoption.
TD Bank's US division is evaluating how AI agents can transform business processes. BNY has deployed "hundreds of digital employees" to handle end-to-end workflows. Commonwealth Bank of Australia hired Ranil Boteju, formerly of Lloyds Banking Group, as its first chief AI officer in early 2026. Lloyds itself appointed Sameer Gupta as chief data and AI officer to lead the bank's AI strategy.
The pattern is consistent: major banks are not experimenting with AI. They are operationalizing it.
What exactly is the FCA worried about?
Mills framed the core issue this way: "As autonomy grows, the nature of regulatory risk changes." When AI recommends and a human decides, responsibility is clear. When AI acts directly, accountability fragments. A customer's AI agent might interact with a bank's AI agent, which might call a third-party AI model. Who owns the outcome?
The review flags third-party dependencies as a particular concern. Banks using external AI model providers, such as OpenAI or Anthropic, must manage not just their own systems but the terms under which external systems access data, initiate actions, and trigger workflows. "Some firms may also allow customer agents or third-party systems to interact directly with their infrastructure," the review notes. That is a new kind of operational risk.
The FCA's response will be to build what it calls an "AI-enabled agentic supervisory model." In practice, this means using AI to supervise AI. Regulators will monitor the transition to autonomous AI, strengthen system-wide oversight, and adapt frameworks as technology evolves.
What should financial services firms do now?
The review offers a direct recommendation: treat governance as a core enabler of AI capabilities, not a compliance afterthought. Firms deploying AI agents need clear accountability structures, audit trails for agent decisions, and defined boundaries for what agents can do autonomously versus what requires human approval.
Managing third-party AI dependencies will require new processes. Banks will need to understand what data flows to external models, under what conditions, and what happens if those models behave unexpectedly. This is not theoretical. A model update from a provider could change agent behavior overnight.
For firms using workflow automation platforms like Zapier, Make, or n8n to connect financial systems, the implications are similar. Any automation that touches financial data now operates in a regulatory environment that is actively tightening. Documentation and audit capabilities matter more than they did a year ago.
Disclosure
Some links in this post are affiliate links — Logicity earns a commission if you sign up, at no extra cost to you. We only link products we have used or actively recommend.
Will other regulators follow the FCA's approach?
Probably. The FCA often moves ahead of other financial regulators, and its frameworks influence European and Commonwealth jurisdictions. The US has been slower on AI-specific financial regulation, though the SEC and OCC have issued guidance on model risk management that touches similar issues.
The challenge for global banks is that regulatory approaches will differ. A system that satisfies UK requirements might not meet US or EU standards. Firms operating across jurisdictions will need flexible governance frameworks that can adapt to multiple regulatory regimes.
Logicity's Take
The FCA's focus on system-wide risk rather than firm-level compliance signals a shift in how regulators think about AI. This is not about whether individual banks use AI responsibly. It is about what happens when the entire financial system runs on interconnected AI agents making decisions at machine speed. Banks that treat AI governance as a competitive advantage, not just a compliance cost, will be better positioned when the rules tighten. The firms that document their AI systems thoroughly now will spend less on retrofitting later.
Frequently Asked Questions
What is an AI agent in banking?
An AI agent is software that can execute tasks autonomously on behalf of a customer or bank, such as making payments, answering queries, or processing applications without human intervention for each step.
When will FCA AI regulations take effect?
The FCA has not set specific compliance dates. The Mills Review outlines a strategic direction for the coming years, with the regulator committing to adapt frameworks as AI adoption evolves.
How are banks using AI agents today?
Major banks are deploying AI agents for workflow automation, customer service, fraud detection, and process optimization. BNY uses hundreds of digital employees for end-to-end workflows, while TD Bank is assessing agent use across business processes.
What risks do AI agents create in financial services?
Key risks include fragmented accountability when AI acts autonomously, third-party dependencies on model providers, potential for system-wide failures if interconnected agents malfunction, and new attack surfaces for fraud.
Need Help Implementing This?
If you're building AI-powered financial workflows and need help with governance frameworks, compliance documentation, or third-party risk management, reach out to our team at Logicity.
Source: Banking Dive
Manaal Khan
Tech & Innovation Writer
Produced with AI assistance and reviewed by the Logicity editorial team. Learn more in our Editorial Policy.
