Meta-backed fellowship puts open source in the firing line
The UK government has launched a $1 million open source AI fellowship funded by Meta through the Alan Turing Institute, inviting the nation’s top AI engineers to spend 12 months building public-sector AI tools.
The scheme focuses on open-source AI models like Meta’s Llama 4 to create applications for everything from speeding up planning approvals to enhancing national security.
Technology Secretary Peter Kyle dubbed it “the best of AI in action – open, practical, and built for public good,” adding that it’s about “delivery, not just ideas – creating real tools that help government work better for people.”
Meta’s Joel Kaplan was also optimistic: “Open-source AI models are helping researchers and developers make major scientific and medical breakthroughs, and they have the potential to transform the delivery of public services too.”
Balancing innovation and risk amid tech dependencies
Yet while open source AI promises transparency, flexibility, and reduced costs, the initiative is not without controversy.
Meta’s involvement raises questions about the true openness of the model and potential commercial interests behind the scenes.
Critics have argued that relying on a company with a history of data privacy issues and market dominance complicates the government’s ambitions for sovereign AI capabilities.
Also, open source models themselves bring challenges like securing sensitive data, ensuring AI outputs are free from bias, and maintaining complex AI systems within public infrastructure are significant hurdles.
The government must tread carefully to avoid simply replacing one form of vendor lock-in with another, especially given ongoing concerns about US tech giants’ influence over UK public services.
This fellowship arrives alongside a high-profile partnership with Google Cloud, which has agreed to provide free cloud and AI services to public bodies including the NHS.
While touted as a boost to efficiency, the Google deal has been criticised by campaigners who warn it risks handing ‘the keys to the data kingdom’ to a US company amid volatile political landscapes.
As London cements itself as a global AI hub second only to San Francisco, the race to secure top AI talent intensifies.
But with 80 per cent of UK AI roles concentrated within the M25, there are warnings from experts like Emma Kendrew of Accenture that this concentration risks creating a two tier tech economy, leaving regions behind in the digital race.
The open source AI fellowship offers a novel way to plug AI expertise directly into government projects, with the hope of unlocking up to £45bn in public sector productivity gains.
Yet success will depend on how well the government balances openness and innovation with security, ethics, and the need to remain independent of dominant tech companies.
Peter Kyle summed up the government’s ambitions: “We’ve already seen the potential. Caddy – developed with Citizens Advice and now helping Cabinet Office teams – shows how open AI tools can boost productivity, improve decision-making, and support frontline staff.”
Whether the fellowship can achieve that vision without falling into new dependencies or unintended risks remains the question at the heart of the UK’s AI push.