The UK government has no shortage of AI strategy documents. There’s the “AI Playbook for the UK Government,” the “AI Opportunities Action Plan,” endless white papers about becoming a “global leader” in responsible AI. Westminster loves a good strategy document.

But here’s the thing: none of it matters much if local authorities can’t actually implement it.

A new piece from Browne Jacobson law firm makes something that’s been obvious to anyone working in government tech worth spelling out: the UK’s AI ambitions are running ahead of its local capacity. The 18% of the UK workforce employed by the public sector directly touches every citizen—but the data held across councils, the NHS, schools, and central government is a mess. Inconsistent formats, no standardisation, unclear ownership. The government’s “National Data Library” ambition sounds great in a press release, but the foundational work hasn’t been done.

There’s also a more subtle tension worth noting. The UK is trying to position itself between the EU’s precautionary approach and America’s dogmatic pro-innovation stance. That’s a reasonable strategy—but it requires sophisticated risk management that most local authorities simply don’t have the expertise to implement. The West Midlands Police incident last year, where AI-generated false intelligence led to a poor decision, is a cautionary tale. Trust in institutions is fragile. When AI screws up publicly, it doesn’t just harm the specific case—it undermines the entire technology’s credibility.

The interesting stuff is happening at the local level. The Waves project in Camden and South Staffordshire councils, collaborating with Google and Demos, is testing how AI could make citizen engagement more meaningful—helping people find consensus on difficult local issues like care services or infrastructure. That’s not glamorous, but it might actually work.

Bottom line: the AI game in Britain won’t be won in Westminster. It’ll be won or lost in town halls and council offices across the country—if they’re given the resources and expertise to actually do the work.