FCA lines up machine-readable rules and API data collection for the agentic age
Jessica Rusu, Chief Data, Information and Intelligence Officer at the UK’s Financial Conduct Authority (FCA), has described how the regulator intends to supervise a financial system in which AI agents increasingly act for customers, and how it plans to become an agentic regulator itself. She was speaking at the Bloomberg UK Policy Series.
Much of the plan rests on data. The FCA opened its Handbook to software through an application programming interface (API) in August, so firms can feed its rules, in machine-readable form, straight into their own AI tools. Rusu said the FCA is now working on machine-readable data collection APIs as well, and that dynamic data collection will form part of the future of supervision. Internally, an AI market first responder agent will supervise around 9,000 small wholesale markets firms that submit transaction reports, and AI tools are being built for the roughly 60,000 anti-money laundering firms due to join the FCA’s remit in 2028.
For agents acting in the market, the FCA plans to work with industry, other regulators and frontier labs on a trusted framework for agentic finance, covering agent identity, delegated authority, accountability and execution. Rusu also argued that significant AI failures and near misses should be reportable.
Machine-readable rules going out and machine-readable data coming in is a loop that increases the efficiency of the software reading requirements and the software meeting them. An agent can only assemble a reliable return if the definitions behind each data point are shared, stable and machine-readable themselves. We might just have a standard or two for regulators going down this path.
Read the full speech here.
