FCA Adds Anthropic Support to Its AI Supercharged Sandbox

The UK Financial Conduct Authority said Anthropic will support the second group of firms in the regulator’s Supercharged Sandbox, a controlled environment for firms experimenting with advanced AI.

The FCA said Anthropic will provide access to Claude, including Claude Code and Claude Cowork, for participants in the sandbox. The second cohort will test AI use cases covering safer agent-led payments and commerce, fraud and economic-crime detection, AI governance, access to financial services for vulnerable and underserved consumers, and compliance or business automation.

The regulator said 21 organisations are joining the second group, including Scottish Widows, Money Advice Trust and TrueLayer. It also said applications rose 51% compared with the first cohort.

The Supercharged Sandbox builds on the FCA’s Digital Sandbox infrastructure from NayaOne and earlier support from NVIDIA. The regulator said the programme is intended to let firms experiment with AI in a safe, controlled environment while relying on existing regulatory frameworks.

Why it matters

For traders and brokers, AI tooling is moving from experimentation into regulated-market workflows. Agent-led payments, fraud detection, governance controls and automated compliance all touch areas that trading platforms and online brokers are likely to build into onboarding, surveillance, customer support and risk operations.

The FCA is not creating a new AI rulebook in this announcement, but it is shaping the testing environment where regulated firms can prove how AI systems behave before broader deployment.

What to watch next

Watch whether sandbox participants turn tests into live FCA-authorised products, especially in payment flows, financial-crime monitoring and compliance automation. Firms using similar AI tools outside the sandbox should also track any FCA expectations that emerge from the cohort.