StandardsAboutContact
BankGenX
UK's FCA Builds Its Own AI 'First Responder'

UK's FCA Builds Its Own AI 'First Responder'

As HM Treasury folds roughly 60,000 more professional-services firms into its anti-money-laundering remit, the Financial Conduct Authority is turning to agentic AI to supervise a perimeter its headcount cannot match alone.

The FCA, under chief executive Nikhil Rathi, is embedding agentic AI into supervision — a 'first responder' screening about a billion rows of wholesale-market data daily — as HM Treasury's October 2025 consolidation folds roughly 60,000 additional firms into its anti-money-laundering remit. The driver is a documented capacity gap, not innovation for its own sake.

BankGenX Desk · 4 min read

The Financial Conduct Authority is deploying agentic AI as a 'first responder' inside its own supervision desk, chief executive Nikhil Rathi said in a 24 June 2026 speech — a shift driven by HM Treasury's October 2025 decision to fold roughly 60,000 additional professional-services firms into the FCA's anti-money-laundering remit. The system screens about a billion rows of wholesale-market data a day. The change is a capacity response, not a technology showcase: headcount is not scaling with the expanded perimeter.

A Single Regulator Inherits a 60,000-Firm Perimeter

HM Treasury, the UK's finance ministry, decided the FCA will become the single anti-money-laundering supervisor for lawyers, accountants and trust and company service providers, retiring the patchwork of roughly two dozen professional-body supervisors that included the ICAEW. The Institute of Chartered Accountants in England and Wales reported in October 2025 that the incoming population runs to about 60,000 additional professional-services firms, layered onto an agency already responsible for the bulk of UK financial-services supervision. Headcount is not scaling to match the expanded remit.

Agentic AI as the FCA's 'First Responder'

Speaking on 24 June 2026, FCA chief executive Nikhil Rathi said the regulator is exploring agentic AI as a 'first responder' screening roughly a billion rows of wholesale-market data a day, working alongside human supervisors to flag potential market abuse faster than manual surveillance allows. Rathi framed the shift around a mismatch he called structural: technology, he said, is moving much faster than regulatory paradigms, and legislation will never keep pace — so supervision itself has to absorb more of the detection burden.

The Compliance Gap the New Perimeter Inherits

The case for automation rests on a compliance baseline, not a hunch. FTI Consulting, citing HM Treasury's AML/CTF supervision data, found only 24% of accountancy firms and 29% of legal firms assessed in 2024-25 were fully compliant with money-laundering rules — 17% and 26% respectively were non-compliant, rising to 28% among the highest-risk legal firms. Those figures, surfacing as supervision was being centralized, are the baseline the FCA inherits: a rate low enough that expanding the population without expanding detection tooling would widen the enforcement gap.

The Same Capacity Strain Shows Up in Washington

The mismatch is not a UK-only story. The FDIC's Consumer Compliance Supervisory Highlights, published in March 2026, report that the agency oversees about 2,755 state-chartered banks but completed only about 825 consumer-compliance exams in 2025, even as its Consumer Response Unit closed 32,128 written complaints and calls — up 21% from 26,451 in 2024. Different regulator, same arithmetic: exam capacity is flat while the supervised population and complaint volume are not, and AI is the tool being reached for to close that gap.

What It Means for Banks, FinTechs and Advisers on the Beat

Firms should not read AI-assisted supervision as lighter-touch. A regulator screening a billion data rows a day, or triaging tens of thousands of new AML files, is more likely to catch undocumented controls, not less. The practical takeaway: AML and market-conduct evidence trails now need to be machine-legible as well as human-readable, because the reviewer on the other side may increasingly be an agent flagging exceptions before a human ever opens the file.

Why is the FCA taking on about 60,000 more businesses under AML supervision?
HM Treasury decided in October 2025 to consolidate UK anti-money-laundering supervision into a single statutory regulator, the FCA, ending oversight by roughly two dozen professional body supervisors, including the ICAEW, which itself reported the ~60,000-firm figure that October.
What does the FCA mean by agentic AI as a 'first responder'?
In a 24 June 2026 speech, CEO Nikhil Rathi described exploring agentic AI systems that screen roughly a billion rows of wholesale-market data a day alongside human supervisors, aiming to flag potential market abuse faster than manual surveillance allows.
Is the FCA's capacity problem unique to the UK?
No. The FDIC's Spring 2026 Consumer Compliance Supervisory Highlights show a similar gap in the US, with only about 825 of 2,755 supervised banks examined for consumer compliance even as complaints rose 21% to 32,128 written complaints and calls.
  1. Watchdogs Build Their Own Agents to Keep Up — PYMNTS
  2. Rethinking regulation for the age of AI — Financial Conduct Authority
  3. Handing AML supervision to FCA will increase costs for business, warns ICAEW — ICAEW
  4. Consumer Compliance Supervisory Highlights, Spring 2026 — FDIC
  5. FCA steps in as single AML watchdog for professional services in the UK — FTI Consulting