On December 17, 2025, the Financial Conduct Authority (FCA) issued a warning regarding a competitive race among major British banks to deploy consumer-facing agentic AI. Financial institutions including NatWest, Lloyds, and Starling confirmed they are working with regulators to trial autonomous AI agents capable of making independent financial decisions for retail customers. These agents are designed to perform proactive tasks such as automatically moving idle cash into higher-yield savings accounts, adjusting investment portfolios in real-time, and setting predictive spending caps based on historical data.
The FCA specifically highlighted that agentic AI introduces risks primarily due to the speed and autonomy at which it operates, which differentiates it from standard generative AI that requires human prompts. Regulators expressed concerns that these agents could react simultaneously to identical market signals, potentially accelerating the pace and probability of bank runs by rapidly shifting large volumes of funds between institutions. Additional risks identified by the watchdog include algorithmic bias in autonomous loan approvals, “AI hallucinations” in financial advisory contexts, and the potential for these systems to prioritize bank profits over customer interests.
To manage these emerging threats, the FCA confirmed it will utilize existing frameworks such as the Consumer Duty and the Senior Managers and Certification Regime to hold banking executives personally accountable for any harm caused by autonomous systems. While the FCA maintains a principles-based approach and does not plan to introduce new AI-specific laws immediately, it has launched live testing initiatives to monitor these trials. Industry forecasts suggest that while 40% of financial firms may adopt AI agents by the end of 2026, a similar percentage of projects could be cancelled by 2027 due to high operational costs and systemic risks.
