The UK FCA’s Supercharged Sandbox for AI in financial services is expanding into its second cohort, with Anthropic providing access to Claude AI models to participating firms.

Twenty-one organisations were selected for the latest cohort, including Scottish Widows, Money Advice Trust, and TrueLayer, reflecting a 51% increase in applications from the first round.

Participants will explore agentic payments, fraud and economic crime detection, AI governance and accountability, and automation of compliance and business operations.

Compliance automation is being tested to handle repetitive regulatory tasks with human oversight remaining in place.

A core focus is AI governance, covering transparency, model oversight, data protection, fairness, cybersecurity, and responsible deployment in regulated settings.

The move signals closer collaboration between regulators and technology firms to shape future AI standards in regulated financial markets.

The FCA intends to rely on existing regulatory frameworks rather than creating new AI rules, aligning with its digital innovation strategy.

The program builds on partnerships with NVIDIA and NayaOne for infrastructure and tools and includes the 10-week Agentic Academy to help firms develop AI agents for finance.

The broader context shows rapid AI adoption across banks, insurers, asset managers, and fintechs, alongside ongoing regulatory, privacy, and resilience concerns.

The program aims to widen access to financial services for vulnerable and underserved consumers.

A key focus is agentic payments, where autonomous AI systems could perform multi-step financial tasks with limited human input, signaling efficiency gains and risk considerations.

The initiative reflects regulatory efforts to assess AI risks and capabilities in the UK financial sector through supervised experimentation and risk mitigation.