Anthropic will support the second group of firms in the FCA's Supercharged Sandbox.
The Sandbox is a controlled environment where firms can safely experiment with advanced AI.
Anthropic will provide access to Claude for participants – including Claude Code and Claude Cowork – to help speed up their development work.
The second group of firms in the Supercharged Sandbox will explore a range of AI use cases. This includes testing solutions designed to:
- enable safer agent-led payments and commerce
- detect fraud and economic crime more effectively
- strengthen AI governance and accountability
- widen access to financial services for vulnerable and underserved consumers
- streamline compliance and business automation
21 organisations, including Scottish Widows, Money Advice Trust and TrueLayer, will join the second group. The selection process followed strong demand, with a 51% increase in the number of applications compared to the first group.
The Sandbox builds on existing support from NayaOne and NVIDIA.
'The high level of interest in the Supercharged Sandbox demonstrates the demand for trusted environments where firms can experiment safely and responsibly,' said Jessica Rusu, chief data, intelligence and information officer, FCA. 'With the support of Anthropic, participants will benefit from the technology they need to accelerate innovation.
'This is central to our commitment to supporting economic growth – enabling firms to make the most of technological advances while maintaining the UK’s position at the forefront of responsible AI adoption and innovation.'
The FCA has also launched the new Agentic Academy, a 10-week specialist AI programme for selected firms which is delivered by the FCA and the Centre for Finance, Technology and Entrepreneurship (CFTE).
Notes to editors
- Firms accepted into the second cohort also include: calQrisk; Merx Digital Solutions Ltd (SmartDrops); Aegis Trace; Sardine AI Corp; Zquas; Trustie Labs; Welleness; IntelXview Ltd; RMI Agentic; Ubyx, partnering with Amazon; Deepflow; FSCom; GAI Labs; Condukt; Kaption; and Relace.
- In 2025, the FCA announced the launch of the Supercharged Sandbox, designed to help firms experiment with AI in a safe and controlled environment. It helps firms who are in the discovery and experiment phase with AI.
- The Supercharged Sandbox builds on the existing Digital Sandbox infrastructure provided by NayaOne, offering advanced compute power to accelerate AI innovation.
- Through a collaboration with NVIDIA, participating firms in the first group were given access to accelerated computing infrastructure and NVIDIA AI Enterprise software to support the development and testing of AI use cases. Firms in the second group will continue to have access to these capabilities.
- The second Supercharged Sandbox received 199 applications in total, compared to 132 in the first iteration.
- The FCA set out how we are working to accelerate digital innovation in our response to the Prime Minister’s letter (PDF), including that we would avoid additional regulations for AI by relying on existing frameworks.
- Read more about how FCA rules apply to AI.
Facts Only
* Anthropic will support the second group of firms in the FCA's Supercharged Sandbox.
* Access to Claude (Claude Code and Claude Cowork) will be provided to participants.
* Use cases explored include agent-led payments and commerce, fraud/economic crime detection, AI governance, widening financial service access, and compliance automation.
* Twenty-one organizations, including Scottish Widows, Money Advice Trust, and TrueLayer, will join the second group.
* The selection process saw a 51% increase in applications compared to the first group.
* The Sandbox builds on support from NayaOne and NVIDIA.
* The FCA launched the Agentic Academy, a 10-week AI programme for selected firms.
* Accepted organizations include calQrisk, Merx Digital Solutions Ltd (SmartDrops), Aegis Trace, Sardine AI Corp, Zquas, Trustie Labs, Welleness, IntelXview Ltd, RMI Agentic, Ubyx (partnering with Amazon), Deepflow, FSCom, GAI Labs, Condukt, Kaption, and Relace.
* The Supercharged Sandbox was announced in 2025 to allow firms to experiment with AI in a safe environment for discovery and experimentation.
Executive Summary
Full Take
The framework establishes a mechanism for channeling high-demand innovation into regulated environments by integrating advanced proprietary AI tools, positioning the regulator—the FCA—as an enabler of responsible technological adoption rather than solely a gatekeeper. The shift from two distinct groups to expanding participation suggests a pattern where controlled experimentation is seen as a prerequisite for mainstream financial AI deployment. The focus on specific use cases (payments, fraud, governance) points toward an institutional desire to mitigate systemic risk while aggressively pursuing competitive technological advantage in the AI space.
The reliance on external providers like Anthropic for access introduces a complex dynamic: firms gain access to cutting-edge tools, and regulators gain oversight through these same channels. This setup tests the boundary between autonomous innovation and mandated accountability. The creation of structured programs like the Agentic Academy alongside the Sandbox suggests a strategy to cultivate an ecosystem where responsible innovation is standardized, rather than simply reacting to risks after deployment. The underlying pattern is the institutionalization of trust through controlled access; success depends on whether these controlled environments can effectively yield solutions that genuinely serve vulnerable consumers and maintain regulatory integrity without stifling the very innovation they seek to govern.
Bridge questions: If the stated goal is responsible AI adoption, what metrics will be used beyond participation numbers to assess the *quality* and *safety* of the experiments conducted in the Sandbox? How does the dependence on third-party foundational models risk creating a regulatory fragmentation if different cohorts benefit from differently curated toolsets? What are the long-term implications for consumer protection when the emphasis is placed equally on streamlining business automation and protecting vulnerable users?
Sentinel — Human
The text functions as a factual report detailing collaborations between the FCA, Anthropic, and various firms within an AI experimentation framework, characterized by structured data and official commentary.