UK FCA Works Toward AI-Enabled Supervision Under Existing Rules

Britain’s Financial Conduct Authority (FCA) is developing an AI-enabled supervisory model to process regulatory data and identify risks earlier, while applying existing consumer-protection and accountability rules to firms’ AI use. It has not set a timetable for broader automation.

At a House of Commons Treasury Committee hearing Tuesday, Oct. 13, FCA Chief Data, Information and Intelligence Officer Jessica Rusu described the current process as keeping “a human in the loop, with every decision.” AI handles unstructured text. The tools also support case officers working in authorization, supervision and enforcement. Rusu said the FCA’s longer-term aim is to automate more work and identify potential harm earlier.

FCA Chief Executive Nikhil Rathi said the regulator receives 1 billion rows of data each week through market surveillance. An early use for its agentic tools is to clean regulatory submissions and flag discrepancies to firms, potentially through automated exchanges. That would let staff focus on higher-risk matters. Rathi also linked market-integrity concerns to round-the-clock trading, tokenization and cross-border capital flows.

A review published July 6 included seven recommendations for the FCA Board and executive to consider. One calls for an AI-enabled supervisory model.

For firms’ use of AI, the FCA plans to rely on existing frameworks instead of creating separate AI rules. The Consumer Duty makes outcomes for retail customers a core obligation for firms. Under the Senior Managers and Certification Regime, senior individuals at regulated firms are accountable for their responsibilities.

A Treasury Committee report published Jan. 20 called for practical FCA guidance by the end of 2026 on applying consumer-protection rules to AI and clarifying accountability for AI-related harm. It also recommended AI-specific stress testing by the FCA and Bank of England.

In an April FCA-commissioned survey of more than 5,000 UK retail financial-services consumers, 20% indicated they would likely use AI capable of acting autonomously toward preset goals. The Oct. 13 hearing set no date for wider automation and did not establish a completed assurance framework for agentic AI. The committee recommended that the FCA publish its guidance by the end of 2026.