The head of the Financial Conduct Authority has put the regulator on record against writing a thicket of new AI rules, arguing that the technology now moves too fast for traditional law-making to matter. Nikhil Rathi told CNBC's Squawk Box Europe that the FCA will not introduce lots of new, detailed rules on artificial intelligence, a stance that sets Britain apart from Brussels at a decisive moment for the industry.
The comments land as adoption reaches near-saturation. More than 80% of financial services firms are already using or trialling AI, shifting the regulator's focus from whether the sector will adopt the technology to how it manages full-scale deployment.
Why the FCA thinks rules cannot keep up
Rathi's core argument is that the rulebook cannot chase a target moving this quickly. He has said legislation will never keep up, pointing out that Buy Now Pay Later regulation took roughly six years to come into force while markets reinvent themselves in months.
The response is a shift in what the FCA sees as its job. Rather than issuing prescriptive rules, it wants to act as a steward of the system, leaning harder on existing frameworks such as the Consumer Duty and the Senior Managers regime, and stepping in through guidance rather than statute. Detailed rules will still apply in some areas, Rathi has said, but not everywhere.
A deliberate break from Brussels
The choice positions the UK against the European Union, whose AI Act imposes prescriptive obligations on systems judged high-risk. Britain is betting that a lighter, collaboration-heavy regime rewards responsible experimentation and pulls in AI investment without gutting market protections.
That gamble carries risk. Some industry voices warn that firms welcoming fewer rules may be poorly prepared for the consequences of looser oversight, and the FCA is reserving the right to tighten expectations if outcomes sour.
The risks the regulator is watching
Rathi has not waved away the dangers. He flagged mounting concentration risk across cloud, model and data providers, noting that 98% of operational incidents reported last year were technology or cyber-related. UK Finance data suggests almost £1.3 billion was lost to payment fraud last year, a figure frontier AI could worsen as attackers gain the same speed as defenders.
Accountability remains the red line. Rathi insists that responsibility for regulated activities must stay clearly assigned no matter how much decision-making is automated, and that consumers will not delegate money to systems they cannot understand. Guidance on good and poor AI practice is due later this year, alongside the Mills Review into AI in retail finance. The rulebook may be thinning, but the scrutiny is not.
