Banks and FinTechs are racing to put AI agents to work. The agencies that police them are doing the same thing. The U.K. Financial Conduct Authority is embedding artificial intelligence across supervision and enforcement as it prepares to absorb about 60,000 additional businesses into its anti-money laundering supervision, a figure the Institute of Chartered Accountants in England and Wales reported in October. That’s on top of the roughly 35,500 firms whose conduct the FCA already regulates.
HM Treasury, the U.K. finance ministry that oversees the country’s anti-money laundering framework, decided the FCA will become the single anti-money laundering supervisor for lawyers, accountants and trust and company service providers, according to the Law Society. That work, ICAEW noted, is now split among roughly two dozen professional bodies and folding it into one agency multiplies the FCA’s caseload overnight. Staff alone can’t cover that jump. The FCA will need software that reads filings, ranks risk and points investigators at the worst offenders first.
In a June 24 speech, FCA CEO Nikhil Rathi said the regulator is exploring agentic AI as a “first responder” to speed up how it monitors wholesale markets, harnessing a billion rows of data per day alongside supervisory judgment to tackle market abuse faster. “Technology is moving much faster than many regulatory paradigms,” Rathi said. The FCA detailed in its 2026/27 annual work program that it will use generative AI to streamline supervision, speed up authorizations and improve how it triages information from firms.
Many of the firms already have significant compliance gaps. FTI Consulting found that only 24% of accountancy firms and 29% of legal firms assessed in 2024-25 were fully compliant with money laundering rules. That gap is exactly why the FCA is leaning on automation instead of headcount alone, and the U.S. is facing the same math.
FDIC Examined Only 825 of Its 2,755 Banks Last Year
U.S. regulators face the same imbalance between supervisory mandates and available staff. The Federal Deposit Insurance Corp. (FDIC) reported in March that it oversees about 2,755 state-chartered banks but conducted only about 825 consumer compliance examinations in 2025, leaving most institutions unexamined in any given year. At the same time, the FDIC’s consumer response unit closed 32,128 written complaints and call records, up 21% from 26,451 in 2024. Complaint volume is rising faster than examination capacity.
The Federal Reserve is running into the same wall. At the end of 2024, it supervised 651 state member banks within community banking organizations and another 43 within regional organizations. Because those institutions are examined on periodic cycles, filings, complaints and transaction data continue to accumulate between visits. Continuous monitoring could close that gap by using AI to assess institutional data as it arrives and direct examiners toward emerging problems, rather than waiting for the next scheduled review.
CFOs Move AI Agents Into Compliance and Cash Management
The shift to agents is running across the economy, and fast. PYMNTS Intelligence found in its report “Agentic AI Breaks Out of the Sandbox” that agentic AI adoption in the services sector jumped fivefold between August and November, from 4.3% to 25%, while adoption at technology firms tripled to 30.8%.
Finance teams are already using agents the way regulators plan to. PYMNTS Intelligence found in its report “CFOs Push AI Forward but Keep a Hand on the Wheel” that nearly half of CFOs use AI to continuously monitor working capital and cash flows. The same technology now flagging risk for finance teams is the technology regulators are racing to master before the gap between them widens.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
