Anthropic joins FCA Supercharged Sandbox with Claude

Anthropic will supply its Claude family of AI models, including Claude Code and Claude Cowork, to firms taking part in the UK Financial Conduct Authority’s second Supercharged Sandbox cohort. The access is intended to support development of AI applications in financial services.

The FCA named 21 organisations in the cohort, including Scottish Widows, Money Advice Trust and TrueLayer. The regulator reported it received 199 applications for the programme, a 51% increase from the first round. Participants plan to use Claude for projects such as agent-led payments, fraud detection, compliance automation and tools for AI governance.

Anthropic’s Claude access will be provided together with computing infrastructure and development tools from NVIDIA and NayaOne. The Supercharged Sandbox offers a controlled setting for teams in the discovery and experimentation phase to test models and new product concepts before moving to live deployment.

At the same time, the FCA opened a 10-week Agentic Academy run with the Centre for Finance, Technology and Entrepreneurship to help firms design and deploy autonomous AI agents for financial tasks. The regulator also operates a separate AI Live Testing programme for firms preparing to use AI systems in live markets.

The FCA established the Supercharged Sandbox to give regulated firms a structured way to trial commercial AI models and partner infrastructure while evaluating operational, compliance and governance issues prior to scaling solutions in production.