A growing race among British banks to implement agentic artificial intelligence (AI) in retail banking is creating new challenges for regulators, particularly around consumer protection, financial stability, and systemic risk. Agentic AI, which can make autonomous decisions and execute tasks without human intervention, is seen as the next frontier in financial technology.

---

What is Agentic AI?

Unlike traditional generative AI, which produces text, images, or code based on human prompts, agentic AI can:

• Plan and make decisions autonomously,
• Execute complex tasks, and
• Learn and adapt based on set objectives.

In banking, this technology promises to revolutionize customer interactions—from automatically moving idle cash into higher-yield accounts to adjusting investment portfolios in response to market fluctuations.

---

British Banks Lead Consumer-Facing Pilots

Leading UK banks, including NatWest, Lloyds, and Starling, are preparing retail customer trials under guidance from the Financial Conduct Authority (FCA). Previously, most agentic AI deployments were limited to back-office operations.

The FCA anticipates that early consumer-facing applications will hit the market in early 2026. According to Jessica Rusu, FCA Chief Data Officer:

“Everyone recognises that agentic AI introduces new risks, primarily because of the ability for something to be done at pace.”

Banks are working within the FCA’s AI sandbox and a live testing initiative designed to balance innovation with regulatory oversight, including frameworks like the senior managers regime and the consumer duty to hold executives accountable and protect customers.

---

Examples of Agentic AI Pilots

• NatWest: Testing AI for accelerated complaints handling, analyzing and resolving cases more efficiently. Full deployment expected in early 2026.
• Lloyds: Piloting tools to help employees assist customers with money management, potentially enabling automatic investments into tax-free ISA accounts.
• Starling: Developing AI tools that create personalized budgets, predictive spending caps, and automated standing orders.

---

Potential Risks of Agentic AI

While agentic AI performs well on simple tasks, experts caution about systemic risks:

• Multiple agents interacting simultaneously could accelerate bank runs by rapidly moving deposits or funds in reaction to identical market signals.
• Reliability concerns arise as AI systems may “hallucinate,” generating plausible but inaccurate advice.
• Advisory limitations: AI may misinterpret user needs, and average bankers may not fully understand the technology’s decision-making process, says Martin Dowdall, lawyer at Taylor Wessing.

Suchitra Nair, head of Deloitte’s EMEA Centre for Regulatory Strategy, warns:

“The systemic risk lies not with one AI agent, but with many interacting and executing actions simultaneously, which could dramatically accelerate financial instability.”

---

Global Perspective

While British banks lead in customer-facing AI pilots, U.S. banks such as JP Morgan currently use agentic AI primarily for back-office operations. Meanwhile, the EU’s AI Act introduces regulatory uncertainty, leaving financial firms in Europe unsure how to deploy agentic AI in retail banking.

Research from Gartner predicts that by 2026, 40% of financial services firms will use AI agents, but more than 40% of agentic AI projects may be canceled by 2027 due to escalating costs and unclear business value.

---

Conclusion

The adoption of agentic AI in British retail banking represents a major technological shift, offering the potential for personalized, automated financial services. However, regulators and banks must navigate complex risks, including systemic financial instability, AI reliability issues, and gaps in executive understanding.

The FCA’s proactive approach, including sandboxes and live testing initiatives, is aimed at ensuring consumer protection while fostering innovation in the rapidly evolving AI-driven banking sector.