The UK's Financial Conduct Authority (FCA) warns of an "arms race" as financial services firms rapidly adopt artificial intelligence. outgoing FCA executive director Sheldon Mills highlighted AI's potential to "democratize" finance, offering sophisticated advice to individuals earning as little as £20,000 a year, previously accessible only to the wealthy.

A report commissioned by Mills recommends the FCA establish an "AI-enabled financial capability service" to provide the public with free guidance on financial choices. Many companies are already piloting AI agents for autonomous financial transactions, but Mills stressed that managers must remain accountable for their AI models' actions, stating, "You need a human on the hook for what they’re doing."

The report also anticipates AI amplifying fraud and cyber threats, citing "deepfakes, synthetic identities, and personalized social engineering." It suggests using AI defensively against these risks. To manage these challenges, the FCA is considering expanded powers under the "critical third parties" regime, potentially supervising major tech providers like Anthropic, Open AI, Amazon, Google, and Microsoft. This could involve stricter disclosure requirements and scenario testing.

These recommendations come amid scrutiny of the FCA's use of AI, including a controversial contract with Palantir to combat financial crime, which some politicians fear could expose sensitive UK financial data to US authorities. The FCA and Palantir deny these concerns.