FCA Expands AI Sandbox as Anthropic Joins to Support Financial Services Innovation

The UK's Financial Conduct Authority (FCA) has expanded its AI innovation programme by bringing Anthropic into its Supercharged Sandbox, providing participating firms with access to the company's Claude AI models to accelerate the development and testing of artificial intelligence applications in financial services.

Anthropic will support the second cohort by offering access to Claude, including Claude Code and Claude Cowork, alongside the existing technology infrastructure provided through partnerships with NayaOne and NVIDIA.

The second cohort comprises 21 organizations, selected from 199 applications, a 51% increase compared with the inaugural programme. Participants include established financial institutions such as Scottish Widows, Money Advice Trust, and TrueLayer, as well as a range of fintech and AI companies.

According to the FCA, firms will use the controlled testing environment to explore practical AI applications across financial services. Key areas of focus include fraud detection and financial crime prevention, AI governance and risk management, agent-led payments and commerce, compliance automation, and expanding access to financial services for underserved and vulnerable consumers.

The Supercharged Sandbox forms part of the FCA's broader strategy to promote responsible AI adoption without introducing AI-specific regulations. Instead, the regulator continues to rely on its existing supervisory and regulatory framework while providing firms with a controlled environment to test emerging technologies.

Alongside the expanded Sandbox, the FCA has also launched the Agentic Academy, a 10-week programme delivered with the Centre for Finance, Technology and Entrepreneurship (CFTE). The initiative is designed to help selected firms develop expertise in deploying agentic AI within regulated financial services.

The latest expansion reflects increasing industry interest in regulatory-backed AI testing environments, with applications for the second Sandbox cohort rising significantly from the first. The FCA said the programme aims to support innovation while ensuring new AI applications are developed in a safe, accountable, and regulated environment.