The head of Britain’s financial watchdog says regulation cannot keep up with advances in artificial intelligence (AI).
Speaking on Wednesday (24 June), Nikhil Rati, CEO of the Financial Conduct Authority (FCA), said regulation needs to evolve to meet the demands of AI use.
“Technology is moving much faster than many regulatory paradigms,” Rati said at the Agents of Change: Generative and Agentic AI in Financial Services 2026 conference. “Legislation will never catch up, which is one reason why the FCA’s strategy is built around rebalancing risk. We are rethinking what it means to be an effective regulator in the age of AI.”
It means considering agent AI as a “first responder” to speed up the way the FCA monitors wholesale markets, leveraging technology and vast datasets, as well as the regulator’s supervisory judgment, to respond more quickly to market fraud.
“In some areas, we will still need detailed rules. But in other areas, traditional rule-making no longer works,” Rati said. “In our role, we will see increased stewardship as well as oversight responsibilities: helping businesses and markets navigate technological change, and working more collaboratively and creatively to understand emerging risks, even acting before the law catches up.”
He likened AI to buy now, pay later (BNPL) companies, saying it took about six years for these companies to enter the FCA’s “regulatory boundary,” but it didn’t take another six years for the agency to act.
Lattice added that the FCA wants to ensure financial services firms can safely build and test AI. In the coming weeks, a Mills Review will be published explaining how AI can reshape retail financial services. And the FCA will be publishing a publication on good and bad AI practices later this year.
Meanwhile, the regulator is offering a supercharged sandbox that allows companies to build and test solutions “using real-world data, computing and systems,” and will work with tech giants such as Nvidia and eventually Google, Rati said.
As PYMNTS wrote earlier this week, sandboxes have limitations. Mudit Gupta, AI practice leader for EY’s U.S. financial services consulting, told Global Finance that most bank leaders believe agent AI can transition more quickly if governance is not seen as a constraint.
“But really, it’s the governance that allows these systems to be deployed at scale,” says Gupta.
Gupta added that although synthetic data is often treated as inherently secure, sensitive signals can be leaked through inference and linking risks. Additionally, he said, replicating and extending historical biases and embedding them behind layers of abstraction makes them more difficult to detect, audit, and challenge.
