The Financial Conduct Authority plans to use AI to automate more supervisory work, including checking firms’ regulatory returns and contacting them when figures do not reconcile. The immediate change is practical: the regulator wants software to handle routine checks so staff can focus on riskier cases, while people retain decision-making authority for now.
Watch Desk analysis
What happened
Resultsense says FCA executive director Jessica Rusu told MPs on 13 October that the regulator wants “to automate wherever we can”. The report says AI is already helping staff process material such as firm correspondence and interview notes, and that an agentic tool has begun contacting firms about inconsistent returns.
According to Resultsense, chief executive Nikhil Rathi said human staff still own every decision “for now”. The article says it draws on reporting by the Financial Times and others; it notes that a published transcript of the committee session was not available.
Why it matters
This is a regulator applying AI to its own oversight, not just setting expectations for firms. If automated checks can flag discrepancies and handle follow-up, supervisors may be able to review more material and concentrate on cases that warrant attention. That is the promise; whether it produces better supervision, rather than simply more automated correspondence, remains to be seen.
The report also describes a wider policy question. Rathi said the boundary for consumer-facing AI tools remains under discussion, particularly where a chatbot moves from general information towards personal recommendations and transactions. That could matter to companies building AI money assistants, as well as to people deciding how much trust to place in them.
Our read
Using agents to tidy regulatory returns is a less glamorous AI story than launching another chatbot, and arguably a more revealing one. It puts the technology inside a consequential public function, where the useful test is whether it helps staff spot problems sooner without obscuring who is accountable. For now, the reported human sign-off is an important boundary, not a permanent guarantee.
What to watch
- What the FCA says in its forthcoming response to the Mills Review.
- Whether the regulator publishes details of the agent’s role, limits and results.
- How the FCA defines the line between general AI information and regulated financial advice.
Discussion spark: Should regulators use AI to automate routine checks on the firms they oversee, or does that risk making supervision harder to challenge?
Sources and evidence
- FCA plans to automate supervision with agentic AI tools – Resultsense (16 October 2026, 08:50 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.