The Financial Conduct Authority has named eight firms for the second cohort of its AI Live Testing programme, marking a significant escalation of the regulator’s hands-on approach to artificial intelligence as British banks prepare to deploy autonomous, agentic systems into customer-facing operations.
In an announcement on 21 April, the FCA confirmed that Barclays, Lloyds Banking Group’s Scottish Widows arm, UBS, Experian, GoCardless and three earlier-stage fintechs — Aereve, Coadjute and Palindrome — will spend the rest of 2026 stress-testing live AI use cases under regulatory supervision. An evaluation report is scheduled for the first quarter of 2027.
Agentic systems take centre stage
The selected use cases reflect where the City believes AI will land first. They include AI-enabled targeted investment support, consumer credit-score insights, anti-money-laundering detection, automated Know Your Customer checks and — most consequentially — agentic payments, in which software agents authorise and execute transactions on a customer’s behalf with limited human intervention.
“We’re continuing to collaborate with firms to support the safe and responsible development of AI in UK financial markets,” Jessica Rusu, the FCA’s chief data, information and intelligence officer, said in the announcement. The programme allows participants to run real-world experiments inside the regulator’s perimeter rather than waiting for formal supervisory clearance, a model that has drawn quiet interest from European and Asian counterparts.
Demand for the FCA’s innovation services has surged. The regulator reported a 49 per cent year-on-year jump in applications to its Regulatory Sandbox and Innovation Pathways, a pace that suggests British compliance teams are no longer treating generative AI as an experimental curiosity.
Supercharged Sandbox prepares second intake
The Live Testing announcement comes a fortnight before the FCA’s separate Supercharged Sandbox opens its second intake on 5 May. That programme, run jointly with NVIDIA, gives selected firms access to GPU compute, curated datasets and a secure cloud environment to develop AI models without breaching data-handling rules.
The first cohort attracted 132 applications and admitted 22 firms, according to FCA figures. For round two, officials have signalled a clear preference for proposals involving agentic AI — including compliance agents, customer-service agents and the same agentic payment infrastructure now being tested in the Live cohort.
The twin programmes form the operational spine of what HM Treasury this month branded a “digital big bang” for UK financial services, an agenda reinforced by the appointment of former FCA executive Chris Woolard CBE as the government’s Wholesale Digital Markets Champion. His brief includes accelerating tokenisation of financial assets, an effort that will sit alongside the AI workstream.
Regulators close ranks
The Digital Regulation Cooperation Forum — comprising the FCA, the Competition and Markets Authority, the Information Commissioner’s Office and Ofcom — published a joint paper this month examining how existing UK frameworks can accommodate agentic systems. The DRCF’s intervention signals that liability, consumer-protection and competition questions raised by AI agents will be addressed across regulators rather than in isolation, a contrast with the more fragmented approach in Brussels and Washington.
For the eight firms in the new cohort, the immediate prize is regulatory cover to deploy AI in production. The longer-term prize, the City hopes, is a British rulebook coherent enough to make London the default jurisdiction for agentic finance — before Frankfurt or Singapore writes one of its own.
Discussion
Sign in to join the discussion.
No comments yet. Be the first to share your thoughts.