London, January 27, 2026, 13:07 (GMT)
- UK government brings in Meta-backed AI specialists to build open-source tools for transport, public safety and defence
- $1 million Meta investment funds a year-long fellowship run through the Alan Turing Institute
- Anthropic to help pilot an optional AI assistant for job seekers later this year
Britain has recruited a team of artificial intelligence specialists, backed by Meta funding, to build AI tools aimed at improving transport, public safety and defence, the government said on Tuesday. (Reuters)
The move lands as ministers try to show quick wins from technology inside government, after launching a separate “CustomerFirst” effort this month to cut long waits and repetitive form-filling in public services.
It also fits a wider push to use AI to raise productivity across government and to keep sensitive work, including on national security, inside secured systems rather than pushing data into outside services.
The Department for Science, Innovation and Technology said a $1 million investment from Meta will fund a new cohort of AI fellows, delivered through the Alan Turing Institute, who will spend the next year developing open-source tools. (Gov)
In transport, the fellows will build models to analyse images and video so councils can better prioritise infrastructure repairs, the department said. They will also develop systems that can run offline or within secured networks for defence and national security teams.
Minister Ian Murray said, “A digital world needs a modern, digital government.” He said the aim is to “re-wire” frontline systems, including in healthcare, policing and transport.
A separate partnership announced on Tuesday will bring in U.S. AI firm Anthropic to help build and pilot an optional assistant for public services, starting with job seekers who need career advice and help finding work. Pip White, Anthropic’s head of UK, Ireland and Northern Europe, said the tie-up would “help deliver on the AI Opportunities Action Plan.”
Meta’s involvement was first flagged in July last year, when the company said the work would use open-source models such as its Llama system. Llama is a “large language model” — software trained on huge datasets that can analyse and generate text, and also work with data such as images, audio and video.
Meta executive Rob Sherman said, “Meta is proud to help bring top British AI talent into government.” Turing’s Mark Girolami said, “AI has huge potential to help us anticipate risks.” (The Alan Turing Institute)
The government said tools built with these models would be owned by the state, letting departments keep sensitive data in-house and adapt the technology for their own needs.
The pitch comes with risks. AI can misread images, amplify bias in decision-making or produce confident but wrong outputs, and public bodies often move slowly on new technology even when pilots go well.
For now, the fellows are set to work over the next year, while the jobseeker assistant pilot is expected to begin later this year, with wider use depending on performance and safeguards.
