The Financial Conduct Authority (FCA) has confirmed that Anthropic will support the second cohort of firms taking part in its Supercharged Sandbox, joining Nvidia in providing technology to 21 organisations working through a discovery and experimentation phase with artificial intelligence. Anthropic will give participants access to its Claude platform, including Claude Code and Claude Cowork, intended to speed up development work across the cohort.
The sandbox is being used to test AI applications spanning safer agent-led payments and commerce, improved detection of fraud and economic crime, stronger AI governance and accountability, wider access to financial services for vulnerable and underserved consumers, and streamlined compliance and business automation. Interest in the programme has risen markedly between cohorts: 199 organisations applied for inclusion in the second group, compared with 132 applications for a similarly sized first cohort, which comprised 11 firms focused on retail innovation and consumer insight and 11 firms focused on wholesale and compliance innovation.
FCA chief data, intelligence and information officer Jessica Rusu said this week that the level of interest showed demand for "trusted environments where firms can experiment safely and responsibly," adding that Anthropic's involvement meant participants would gain the technology needed to accelerate innovation. She described the initiative as "central to our commitment to supporting economic growth," framing it alongside the regulator's efforts to keep the UK at the forefront of AI adoption.
Anthropic's participation follows an episode earlier this year, in April, when the company said its Claude Mythos tool could outperform humans at some hacking and cybersecurity tasks, a disclosure that generated political and financial concern at the time. Nvidia, meanwhile, is continuing an existing relationship with the sandbox: the chip maker, which became the world's most valuable company by market capitalisation in 2024, supplied accelerated computing infrastructure and Nvidia AI Enterprise software to the first cohort, and second-cohort participants will also have access to its tools.
Infrastructure for the sandbox continues to rest on a separate arrangement with NayaOne, the London-based fintech company that was awarded a three-year contract worth more than £1m (about $1.26m) to run the FCA's digital sandbox on a permanent footing. Rusu said in an April speech that the authority was extending its partnership with NayaOne, which underpins the newer Supercharged Sandbox alongside the original digital sandbox. The Supercharged Sandbox sits within the FCA's wider AI Lab, which also encompasses an AI live testing proposal published last year and a separate cohort of firms already live-testing AI applications.
The second Supercharged Sandbox cohort runs until 31 December 2026, with a showcase demonstration day scheduled for 26 November, setting a fixed window within which participants' work on Claude and Nvidia infrastructure covering payments, fraud detection and governance use cases will be brought forward for public presentation.
