U.K. Regulator Warn of Agentic AI Risks as British Banks Lead Adoption
In Focus
- The FCA supports banks to safely experiment AI
- British Banks Lloyds and NatWest are preparing for AI agent trials
- Regulators say AI agent autonomy could impact financial stability
The Financial Conduct Authority (FCA) has warned that adoption of agentic AI by British banks
could create new risks for customers due to their ability to make decisions and act on their own. According to Yahoo Finance, the financial watchdog has committed to ensuring that banks don’t overlook customer interests as they rush to adopt agentic AI.
British Banks Take Lead in Agentic AI
Financial institutions in the U.K. are leading European banks in testing AI agents for customer service. This is driven by FCA’s approach towards innovation, which includes setting up an AI sandbox to support early AI agents’ customer banking trials.
The financial watchdog partnered with NVIDIA to supercharge the AI sandbox, enabling banks to experiment safely with AI. The FCA also introduced a live-testing program for companies that are ready to launch AI tools. The EU’s AI act has created uncertainty regarding agentic AI in the finance sector. The law is aimed at fostering responsible AI development and deployment in the region.
Forecasts by research firm Gartner show that 40% of financial service firms will be using AI agents by the end of 2026. Gartner has also warned that over 40% of agentic projects across industries will terminate by the end of 2027 due to rising costs and insufficient risk controls.
How Agentic AI Could Change the Financial Sector
Agentic AI could shift how customers budget, save, and invest. Unlike generative AI, which creates images, code, or text, agentic AI comes with planning, decision making, learning, and task implementation capabilities that are designed to achieve specific goals.
As such, banks can customize AI agents to adjust customer investments when markets shift or move unspent funds to higher-interest accounts. In the U.S., banks such as J.P. Morgan have rolled out agentic AI to perform back-office tasks. In the U.K., banks like Lloyds and NatWest are working with the FCA to prepare for retail-customer trials.
According to the Financial Conduct Authority, agentic AI could become part of customer-facing applications in the banking sector early 2026. Already, NatWest has commenced AI agent trials to fast-track complaint handling. Last month, Lloyds announced plans to test the technology on its employees with the aim of helping customers manage their funds better.
Agentic AI Risks in the Financial Sector
Regulators and financial experts view autonomy and AI’s high execution speeds as a huge risk to financial governance and stability. Recently, Zoho AI agent raised governance concerns after leaking business secrets during a startup interaction.
“Everyone recognises that agentic AI introduces new risks, primarily because of the ability for something to be done at pace,” FCA Chief Data Officer Jessica Rusu said as stated by Yahoo Finance.
Experts also warn that AI agents can fail in performing complex tasks despite their efficiency in completing simple, short tasks. They say systemic risks emerge when many agents interact and execute actions simultaneously.
As part of FCA’s agentic AI regulation, the watchdog plans to apply existing rules to hold senior managers accountable for misconduct and ensure customer interests come first.
