UK Financial Conduct Authority Selects 21 Firms for Second AI Supercharged Sandbox Cohort with Anthropic Support

The Financial Conduct Authority (FCA), the United Kingdom’s primary financial services regulator, has officially announced the selection of 21 pioneering organizations for the second cohort of its highly anticipated "Supercharged Sandbox." This initiative, designed to push the boundaries of financial technology within a safe regulatory environment, will pivot its focus toward advanced artificial intelligence (AI) applications. Specifically, the second group will explore "agentic" AI use cases, focusing on autonomous agent-led payments, enhanced fraud detection, and the broadening of financial inclusion for underserved populations. In a significant strategic move, the FCA also revealed a new partnership with the AI safety and research company Anthropic, which will provide participants with high-level access to its Claude AI models to accelerate development.

The Supercharged Sandbox represents an evolution of the FCA’s original regulatory sandbox, which launched in 2016 as a global first. While the original iteration allowed firms to test innovative products with real consumers, the "Supercharged" version provides an enhanced technological layer, offering participants access to high-performance computing (HPC) and sophisticated software tools. The announcement of the second cohort comes amid a surge of interest in generative AI and autonomous systems, reflecting a broader trend in the global financial sector to integrate "agentic" capabilities—AI that can not only analyze data but also execute complex tasks and transactions on behalf of users.

Strategic Focus Areas and Selected Participants

The 21 organizations selected for this cohort represent a diverse cross-section of the financial ecosystem, ranging from established insurance giants and fintech innovators to non-profit advisory bodies. Notable participants include TrueLayer, a leader in open banking infrastructure; Scottish Widows, the long-standing life insurance and pensions provider; and the Money Advice Trust, a charity focused on helping people across the UK tackle debt.

The cohort will focus on five primary pillars of AI experimentation:

1. Agent-Led Payments and Commerce: Developing AI agents capable of negotiating and executing transactions autonomously or semi-autonomously, streamlining the path from intent to purchase.
2. Fraud Detection and Economic Crime: Leveraging machine learning to identify sophisticated criminal patterns and prevent money laundering in real-time.
3. AI Governance and Accountability: Strengthening the frameworks that ensure AI decisions are transparent, ethical, and attributable to human oversight.
4. Financial Inclusion: Using AI to assess the needs of vulnerable or "thin-file" consumers who are often excluded from traditional credit and banking services.
5. Compliance and Business Automation: Streamlining the often-burdensome regulatory reporting processes and internal administrative tasks through intelligent automation.

Chronology of the FCA’s Innovation Framework

The trajectory of the FCA’s sandbox initiatives demonstrates a clear commitment to maintaining the UK’s status as a global fintech hub. The timeline of this development reveals a steady progression from basic experimentation to deep-tech integration.

• 2016: The FCA launches the world’s first regulatory sandbox, providing a framework for firms to test innovative business models without immediately facing the full weight of regulatory requirements.
• 2021-2022: Following the Kalifa Review of UK Fintech, the government and the FCA recognize the need for more permanent, data-driven testing environments.
• 2023: The "Supercharged Sandbox" is launched. This new model moves beyond simple regulatory relief to provide actual technical infrastructure. The first cohort focused heavily on the initial wave of generative AI and data analytics, supported by NVIDIA’s accelerated computing resources.
• Late 2024: Applications open for the second cohort. The FCA reports a significant uptick in interest, receiving 199 applications—a 51% increase from the 132 applications received for the first cohort.
• Early 2025: The FCA announces the 21 selected firms and the inclusion of Anthropic as a key technology partner, signaling a shift toward agentic AI and LLM-driven solutions.

The Role of Anthropic and Infrastructure Partners

A cornerstone of the second cohort’s value proposition is the technical support provided by industry leaders. While the first cohort benefited from NVIDIA’s hardware and AI Enterprise software, the second cohort will see an expanded ecosystem. Anthropic will provide participants with access to "Claude Code" and "Claude Cowork." These tools are designed to assist developers in writing, debugging, and deploying code more efficiently, effectively acting as an "AI pair programmer" for the fintech engineers within the sandbox.

The FCA’s Jessica Rusu, Chief Data, Intelligence, and Information Officer, emphasized that this partnership is about more than just providing tools; it is about "accelerating innovation" in a way that remains "responsible." By using Claude, a model known for its focus on "Constitutional AI" (AI guided by a set of ethical principles), the FCA is subtly signaling its preference for safety-first development.

Furthermore, the digital infrastructure of the sandbox continues to be powered by NayaOne. As a fintech ecosystem provider, NayaOne offers a "Digital Sandbox" environment that includes synthetic data sets, which are crucial for testing AI without compromising the privacy of real UK citizens. This environment allows firms to simulate market conditions, consumer behaviors, and even cyber-attack scenarios to see how their AI models hold up under pressure.

Data and Market Context: The Surge in AI Demand

The 51% increase in applications for the Supercharged Sandbox is not an isolated metric; it mirrors a broader explosion in AI investment within the UK. According to recent market data, the UK AI sector is currently valued at over $92 billion and is projected to contribute significantly to the national GDP over the next decade. By providing a "trusted environment," the FCA is addressing one of the primary hurdles to AI adoption in finance: the fear of regulatory backlash or unforeseen systemic risks.

Karan Jain, Founder and CEO of NayaOne, noted that the high volume of applications—nearly 200—proves that firms are eager to move past the "hype" phase of AI and into the "proof of concept" phase. The demand for agentic AI, in particular, suggests that the industry is looking for ways to reduce operational costs and improve customer outcomes through high-level automation.

Official Responses and Industry Perspectives

The FCA’s leadership has framed this initiative as a vital component of the UK’s economic growth strategy. Jessica Rusu stated, "The high level of interest in the Supercharged Sandbox demonstrates the demand for trusted environments where firms can experiment safely and responsibly. With the support of Anthropic, participants will benefit from the technology they need to accelerate innovation." She further noted that this is central to maintaining the UK’s position at the "forefront of responsible AI adoption."

From the private sector, the sentiment is equally optimistic. Karan Jain of NayaOne highlighted the "global leadership" of the FCA, suggesting that other international regulators are looking to the UK’s sandbox model as a blueprint. Industry analysts suggest that by bringing in a non-profit like the Money Advice Trust alongside commercial giants like Scottish Widows, the FCA is ensuring that AI development does not just benefit the bottom line of large corporations, but also provides a "social dividend" by addressing debt and financial exclusion.

Analysis of Implications: Toward an Autonomous Financial Future

The shift toward "agentic" AI within the sandbox has profound implications for the future of financial services. Traditional AI in banking has been largely reactive—detecting a fraudulent transaction after it happens or providing a chatbot response to a query. Agentic AI, however, is proactive.

For a firm like TrueLayer, agentic AI could mean developing systems where a user’s AI assistant negotiates a better utility rate and switches providers automatically, handling the payment and identity verification without the user needing to fill out a single form. For Scottish Widows, it could mean AI agents that constantly monitor a customer’s pension portfolio and suggest (or execute) rebalancing strategies based on real-time economic shifts and the customer’s changing risk profile.

However, this level of autonomy brings significant regulatory challenges. If an AI agent makes a financial error, who is liable? The developer of the AI, the firm that deployed it, or the infrastructure provider? By testing these scenarios in the Supercharged Sandbox, the FCA can develop "proportional" regulations that address these risks without stifling the technology in its infancy.

Broader Impact and the Roadmap to 2030

The Supercharged Sandbox is a key tactical element of the FCA’s broader "Open Finance Roadmap to 2030." This roadmap envisions a future where data flows securely across all financial products—not just banking, but also mortgages, pensions, and insurance. AI is the "engine" that will process this data to provide personalized financial advice and services to millions of people.

Furthermore, the collaboration with Anthropic and NVIDIA positions the UK as a bridge between the rapid technological advancement of Silicon Valley and the stringent regulatory standards of Europe. While the European Union’s AI Act takes a more prescriptive, "risk-based" legislative approach, the UK’s sandbox model offers a more "outcome-based" and collaborative path. This could make the UK a more attractive destination for global fintech firms looking to scale AI products in a regulated yet supportive environment.

As Cohort 2 begins its work, the financial industry will be watching closely. The outcomes of these tests will likely inform future FCA rulebooks and could set the standard for how AI agents are integrated into the global economy. The success of this cohort will be measured not just by the technological breakthroughs achieved, but by the establishment of a "gold standard" for AI safety and governance in one of the world’s most complex and vital sectors.