AI Safety4 min read

FCA warns of AI 'arms race' in financial services, urges powers

Sheldon Mills urges a three-to-six month review of ChatGPT, Claude and Gemini use in finance and calls for new supervisory powers over Big.

The Brieftide

TL;DR

  • 01Sheldon Mills urges a three-to-six month review of ChatGPT, Claude and Gemini use in finance and calls for new supervisory powers over Big.
  • 02It also suggests the FCA could seek extra powers under the "designated activities regime" to regulate specific activities without authorizing the firms that carry them out.
  • 03Mills recommends the FCA convene public and private groups to build an "AI-enabled financial capability service" offering free guidance on personal financial choices.

The Financial Conduct Authority warned regulators are in an "arms race" to keep up with artificial intelligence in financial services, and Sheldon Mills urged the regulator to seek new powers and a formal review of large language models such as ChatGPT, Claude and Gemini.

Mills, an FCA executive director, set out those recommendations in a report he commissioned on AI in financial services, telling the FT the watchdog must embrace AI itself to "monitor, detect, and tackle the risks." He said the regulator should examine whether the use of large language models falls within its rules.

What are the FCA's main recommendations?

The report urges a review within three to six months and calls for expanded supervisory powers, including using the "critical third parties" regime to supervise providers such as Anthropic, OpenAI, Amazon, Google, and Microsoft. It also suggests the FCA could seek extra powers under the "designated activities regime" to regulate specific activities without authorizing the firms that carry them out.

Mills recommends the FCA convene public and private groups to build an "AI-enabled financial capability service" offering free guidance on personal financial choices. The report also proposes boosting powers to supervise key technology providers, including imposing disclosure requirements, annual self-assessments and scenario testing of resilience to severe disruptions.

How widespread is consumer appetite for AI financial advice?

Research commissioned by Mills found a fifth of UK adults were open to using AI models to make financial decisions for them, such as on savings or borrowing, even though those models are not covered by regulation and offer no recourse to compensation if things go wrong. The report flags that some firms view AI-based advice as an "economically equivalent" service that sits outside the regulatory perimeter, despite existing rules for regulated companies giving similar recommendations.

Mills highlighted potential consumer benefits as well, saying AI could "democratize" finance by widening access to sophisticated services, for example giving someone earning £20,000 a year access to advice usually available only to "somebody who has got £10mn in savings or assets." He also warned, "It is an arms race."

What risks did the report identify?

The report lists several risks: hyper-personalization enabling bias, opaque pricing, and personalized manipulation; an increase in fraud and cyber attacks through deepfakes, synthetic identities and personalized social engineering; and the challenge that many firms are piloting AI agents that can autonomously carry out financial transactions. It also stresses that managers must remain accountable, noting "You need a human on the hook for what they're doing."

The paper calls for AI to be used defensively to combat fraud and cyber threats while recognizing the technology will likely "amplify" those threats. It recommends stronger oversight of technology providers whose services are critical to the financial sector.

Why it matters

If the FCA acts on Mills' recommendations, the regulatory perimeter for financial advice and services could expand to cover interactions driven by large language models, shifting where liability and consumer protections sit. Reining in or supervising large AI providers would also force closer engagement between regulators and major cloud and AI companies named in the report. That would change how consumer-facing finance products are built and who can be held accountable when automated advice goes wrong.

What to watch

The FCA board is due to discuss Mills' report before deciding how to respond. Watch for a formal review launched within the next three to six months and any government decision to designate Big Tech firms under the critical third parties regime, which would trigger tougher disclosure and resilience testing.

Additional facts: Mills is leaving the FCA after eight years; the watchdog faced scrutiny over a 12-week contract with Palantir to test AI for fighting financial crime, a contract the FCA and Palantir say does not give US authorities access to sensitive UK financial information. The report recommends convening public and private sector groups to create an AI-enabled financial capability service for the British public.

Advertisement

Written by The Brieftide · Source: Ars Technica

The Brieftide Daily · 06:00

Briefs like this one, in your inbox every morning.

 

FreeOne email a dayEvery claim sourcedUnsubscribe in one click

Continue reading

More in AI Safety
Advertisement