13 November, 2026 | 4 min read

What does it cost local government to do nothing about AI?

If we demand evidence of the ROI from AI projects, we should also demand evidence of the cost of doing nothing. Councils in England face a projected £7 billion funding gap by 2028/29, according to LGA analysis, alongside rising demand and workforce pressures. That gap is the starting point for any decision about AI, including the decision to wait. 

I recently spoke to an audience of local government CFOs and Section 151 officers about AI. Their scepticism was fair, and one argument stood out in particular: councils spend more time handling AI-generated complaints and FOI requests than they save by adopting AI themselves. 

And there’s some truth in that. Even the ICO recognises the growing pressure from AI generated FOI requests. It published guidance in May 2026 for public authorities dealing with higher volumes of them. 

That demand is a real cost, and councils can respond with the same technology. AI-assisted triage can categorise incoming requests, identify duplicates, locate relevant records and draft responses for officer review. A resident using ChatGPT to write a complaint is a demand management problem. A designed solution (built on trusted data with appropriate security, clear controls and human oversight) is a different thing and it can mitigate the first problem. 

Each of the four main AI risks has a familiar control 

Hallucinations, bias, security and auditability are legitimate concerns for local government finance professionals, whose careers rest on auditable evidence, accountability and managing risk. Each has a control that council governance already uses. Officers sign-off and check outputs before anyone acts on them, and grounding answers in council-held data limits invented content. An equality impact assessment tests for bias, a data protection impact assessment addresses security and data protection, and logged inputs and outputs provide the audit trail.

Co-creation supports decisions and leaves them with people

I also introduced the concept of AI co-creation to the audience. Imagine bringing service leaders, finance professionals and technologists together to explore a business problem, with AI actively participating in the conversation, drawing on wider evidence, challenging assumptions and suggesting options.

Take an overspend on care placements, for example. AI could help the group ask:

Placement costs

Which placements outside the area cost the most, and what drove each decision?

Cost comparisons

Where do comparable councils achieve similar outcomes at lower cost?

Hidden costs

Which cost drivers sit outside current budget reports?

One CFO heard my example as a claim that AI could solve the budget problem. But this was a misunderstanding of the role of AI, so to be clear: AI does not make the decision. It helps us ask better questions, examine more evidence and consider options we might otherwise overlook. Expertise, judgement and accountability stay with the people in the room. 

Are we so focused on what AI might get wrong that we’re failing to understand what it could help us get right? 

Evidence from councils so far 

At Kingston Council, an early AI pilot reported a 50 to 60% reduction in time spent producing social care case notes and assessments. The gain that matters most is time returned to social workers: more time with vulnerable people and better use of scarce professional expertise. 

The LGA reports that 95% of councils responding to its AI survey were using or exploring the technology, but only 7% were considered innovative leaders. Staff capability, capacity and funding remain the main barriers. 

Government estimates suggest over £45 billion a year in potential public sector productivity benefits through digitisation, including AI. These are estimates, not guaranteed savings. 

Imagine what could be achieved if we focused not simply on saving money, but giving social workers more time with vulnerable people, improving the support available to families and making better use of scarce professional expertise.

AI should enhance human capability, not replace human accountability

My concern is that local government risks treating caution as a strategy, when doing nothing carries risks of its own.

CFOs and Section 151 officers are right to demand a robust business case, demonstrable outcomes and clear financial accountability for AI investment.

The same scrutiny should apply to the decision not to act. Three steps put that into practice:

Process baseline

Baseline the time and cost of one high-volume process.

AI pilot

Pilot an AI solution against that baseline, with a measured return.

Inaction costs

Add a “cost of inaction” line to every business case.

Run those steps and the cost of doing nothing becomes a figure the council can weigh against the cost of acting.

Discover how Version 1 helps public sector organisations harness AI, modernise services and improve operational efficiency. From AI strategy and implementation to data-led transformation, we help councils deliver measurable outcomes while maintaining governance, security and human accountability.