Mark Zuckerberg Chairman and CEO of Meta Platforms (formerly Facebook, Inc.) | Meta Platforms (formerly Facebook, Inc.)
Meta is providing $1 million in funding for an initiative aimed at leveraging the expertise of the UK's top AI talent to enhance public services using open source models, including Meta's Llama models. The Open Source AI Fellowship seeks to address a common governmental challenge: integrating innovation and technical expertise from the private sector into government operations.
The initiative emphasizes open source AI as crucial for the UK to build its sovereign AI capabilities. This approach allows the government to develop its own AI tools independently of foreign closed systems. Open source AI models have already facilitated significant scientific and medical advancements and hold potential for transforming public service delivery.
With Meta's financial backing, the fellowship will be managed by the Alan Turing Institute (ATI). It will place AI experts in various government departments to tackle significant challenges openly and in the public interest. Potential projects include developing AI tools for secure applications like language translation in national security and utilizing construction planning data to expedite housing approvals. Fellows may also contribute to expanding "Humphrey," a suite of AI tools designed to alleviate administrative burdens on civil servants.
A key aspect of this program is that any solutions developed through it remain government-owned, ensuring sensitive data stays within governmental control and allowing for adaptation without reliance on proprietary systems. The use cases created will be made available as open source for broader public use.
Meta recently collaborated with British think tank Social Market Foundation on a report indicating that treating open source AI as a viable option could provide better taxpayer value, increased strategic autonomy, enhanced security auditing, and greater public legitimacy.
The fellowship program represents a step towards realizing these benefits. By focusing on open source AI models, it aims to reduce taxpayer costs and unlock up to £45 billion in productivity gains across the public sector. The fellows' work in areas such as national security, public health, and planning reform is expected to ensure more immediate and direct results from public investments.
The fellowships are set to commence in January 2026 and will last 12 months, with applications opening soon.
"We hope these fellows will make a big, positive difference and help to demonstrate just how valuable open source AI can be to governments and society more broadly."