Anthropic joins the FCA's Supercharged Sandbox to accelerate financial AI testing as the FCA's Mills Review finds 11 million UK adults willing to use autonomous AI agents in financial services

Anthropic, the artificial intelligence research company, has been confirmed as a participant supporting a second cohort of companies in the Financial Conduct Authority's Supercharged Sandbox, an experimental regulatory environment the FCA created to allow accelerated testing of advanced AI technologies in financial services. The announcement was reported by Law360. Separately, the FCA's Mills Review, published in July 2026, found that approximately 11 million UK adults would be willing to use AI capable of acting autonomously on their behalf within predefined goals, even as trust and control remain significant concerns for many consumers. The review identifies a shift in how AI is being deployed: from tools that advise humans to agents that act for them, executing payments, switching products, and taking positions before a human has an opportunity to intervene. Together, the Anthropic sandbox participation and the Mills Review findings frame a significant regulatory and legal moment. The FCA is now grappling with a fundamental accountability question: when an AI agent causes financial harm by executing an authorised action incorrectly, who bears regulatory responsibility? The consumer, the firm deploying the agent, the model provider such as Anthropic, or some combination of all three? For legal practitioners, these questions engage financial services regulation, consumer protection law, and AI liability frameworks simultaneously. The agentic finance context also raises specific issues around the scope of what an agent is authorised to do on behalf of a client, the adequacy of consent frameworks, and how existing financial promotions and suitability rules apply when decisions are made by an algorithm rather than a human adviser.