The FCA announces that Anthropic has joined the second cohort of the AI Super Sandbox, with application volume up 51% year over year

The UK Financial Conduct Authority (FCA) announced on July 23 that the second phase of its “Supercharged Sandbox” program has officially launched. Anthropic will provide Claude AI models to 21 participating organizations, including Claude Code and Claude Cowork. A total of 199 applications were received for the second phase, up 51% from the first round.

FCA Second-Phase Supercharged Sandbox Technology Partnership Framework

According to the FCA announcement, the technical collaboration framework for the second phase of the supercharged sandbox is as follows:

Anthropic: Provides Claude AI models (including Claude Code and Claude Cowork) for use by 21 participating organizations

NVIDIA: Provides computing infrastructure

NayaOne: Provides development tools

Application areas: Agent-led payments, fraud detection, compliance automation, AI governance

Second-Phase Applications Hit 199, Up 51%

According to the FCA’s announcement on Wednesday, the second phase received 199 applications in total, up 51% from the first round. The second cohort includes 21 institutions, including Scottish Widows, Money Advice Trust, and TrueLayer.

The FCA’s program has two tiers: the Supercharged Sandbox is designed to help companies in the AI development exploration and experimentation stage test applications in a controlled environment. The FCA’s separate AI live testing program supports companies that are ready to use AI in real markets.

FCA Also Launches 10-Week Agentic Academy

In addition to the second phase of the Supercharged Sandbox, the FCA, in collaboration with the financial, technology, and entrepreneurship center, simultaneously launched a 10-week Agentic Academy. The initiative aims to help businesses develop and deploy AI agents (AI Agents) in the financial services sector.

FAQ

What tools does Anthropic provide in the FCA’s second phase of the supercharged sandbox?

According to the FCA’s July 23, 2026 announcement, Anthropic provides Claude AI models, including Claude Code and Claude Cowork, to the 21 participating organizations in the second phase of the supercharged sandbox. The application scope includes agent-led payments, fraud detection, compliance automation, and AI governance.

How much did the application volume increase for the FCA’s second phase of the supercharged sandbox?

According to the FCA announcement, the second phase received 199 applications, up 51% from the first round, reflecting continued growth in demand from financial companies for controlled AI testing environments.

How does the FCA supercharged sandbox differ from the AI live testing program?

According to the FCA, the supercharged sandbox is for companies in the AI development exploration and experimentation stage, which test in a controlled environment. The FCA’s separate AI live testing program supports companies that are ready to deploy AI in real markets.