Anthropic joins FCA Supercharged Sandbox for AI experimentation
Anthropic will provide Claude tools to the second cohort of FCA Supercharged Sandbox firms testing AI use cases including agent-led payments.
The FCA is deepening its AI supervisory engagement by partnering with frontier model providers. The sandbox will explore agent-led payments and commerce, indicating regulatory openness to autonomous AI in financial services while shaping future guardrails on model risk, accountability, and consumer protection.
Signals FCA direction on AI governance, agentic payments, and model risk expectations for regulated firms.
Action Required
Monitor sandbox outputs and align AI governance frameworks with emerging FCA expectations on agentic AI and payments.
Shapes upcoming UK AI supervisory approach and sets precedent for agentic AI in regulated activities.
Update AI risk taxonomy to include agentic AI and third-party model dependencies. Track sandbox learnings to pre-empt FCA expectations on AI accountability, testing, and consumer outcomes under Consumer Duty.
“Anthropic will support the second group of firms in the FCA's Supercharged Sandbox, a controlled environment where firms can safely experiment with advanced AI, providing Claude access to test agent-led payments and other use cases.”
Published: 2026-08-02