The UK government has announced a new $1 million Open-Source AI Fellowship, inviting the country’s top AI engineers to help transform public services using open-source models like Meta’s Llama 3.5. Backed by a Meta grant to the Alan Turing Institute, the 12-month program begins in January 2026 and aims to unlock AI-driven innovations across government operations.
Fellows will work within the Department for Science, Innovation and Technology’s (DSIT) Incubator for AI—the team behind “Humphrey,” an AI toolkit for civil servants—and collaborate with departments on high-impact challenges, such as speeding up construction approvals, enhancing national security translation tools, and expanding AI systems that support ministers’ decision-making.
The initiative follows the successful deployment of “Caddy,” an AI customer support assistant co-developed with Citizens Advice, which has halved response times and improved staff confidence across six UK call centres. The tool, now open-sourced and in use by the Cabinet Office, provides quick, accurate information on grant decisions and citizen guidance topics.
Technology Secretary Peter Kyle called the Fellowship “the best of AI in action—open, practical, and built for public good.” Fellows will focus on real-world use cases and build tools entirely open to the public sector, with potential to reduce AI costs and unlock up to £45 billion in productivity gains.
Applicants can register interest now, with full applications opening soon. In tandem with the Fellowship, the government is launching the next phase of its AI Knowledge Hub, offering tools, prompt libraries, and best practices to help public servants scale AI use effectively.
This marks a major step in the UK’s ambition to use AI not just for efficiency, but for equitable, secure, and accessible public service delivery.
Need Help?
If you have questions or concerns about any global guidelines, regulations and laws, don’t hesitate to reach out to BABL AI. Their Audit Experts can offer valuable insight, and ensure you’re informed and compliant.
