Spotlight on Policy Tech and Regulation
Artificial Intelligence
13 June 2025
Local government is adopting AI, but careful thought is needed
Artificial intelligence is becoming a common tool across local government services. From chatbots handling customer inquiries to automating paperwork and addressing housing issues like damp, AI is already in use. Central government is piloting AI tools designed for local authorities, while some councils partner with commercial AI providers to explore different applications.
Somerset Council was an early adopter, joining Microsoft’s Early Access Program for AI. They began with Copilot, a generative AI chatbot, and now use Magic Notes, which converts meeting recordings into initial drafts of documents. This tool is used by about half of the UK's social care teams. It has improved paperwork quality and halved the time staff spend on documentation, freeing frontline workers to focus more on person-centred tasks and improving their work-life balance.
According to Alexander Iosad, director of government innovation policy at the Tony Blair Institute, AI offers significant opportunities to enhance democratic processes and decision-making. AI can engage citizens, streamline data collection, accelerate back-office tasks, and improve policy decisions. Traditional policymaking often struggles with excess data that can’t be translated into actionable insights. AI can process large volumes of data quickly, keeping information current and cross-referencing it across services for better outcomes.
Despite the potential, challenges remain. A survey by the County Councils Network (CCN) showed that 85% of county and unitary councils are already using AI, with the rest considering it. However, many cite staff capacity, funding, and training as barriers. The Tony Blair Institute highlights that councils often lack the confidence, skills, and infrastructure to fully benefit from AI innovation.
Working with one council, the institute estimated AI could automate about 26% of tasks, saving one million work hours or £30 million annually. If scaled nationally, this could translate to around £8 billion in savings. While organisations, especially public services, seek cost reductions from AI, less than half of councils surveyed have seen direct savings so far. However, over 90% report improved staff productivity and three-quarters say services improved.
Major digital transformations require investment to deliver benefits. Public sector projects have a history of cost overruns and delays. For example, Birmingham City Council’s £19 million IT system investment ballooned to £90 million before becoming operational.
There are also concerns about how some AI systems are used. Several councils deploy AI to identify families needing support, spot young people at risk of gang involvement, or detect council tax fraud. These uses have faced criticism over bias, consent, and transparency.
In Rotterdam, an algorithm scored residents' fraud risk based on data from previous investigations. It disproportionately flagged women, young people, parents, and migrants. Subjective caseworker assessments were embedded in the data, creating a false impression of neutrality. The city suspended the system in 2021 after an ethics review.
In 2015, Hackney Council used the Early Help Profiling System by Xantura to identify families at risk without informing them their data was used in that way. These examples underline the need for clear governance, transparency, and privacy safeguards in AI applications within local government.
The Local Government Association advises councils to manage AI risks carefully, focusing on strong data foundations, data protection, bias prevention, and privacy.
The Tony Blair Institute recommends councillors support AI innovation while managing risks by adopting proven tools, collaborating with partners, applying existing standards, and improving workforce AI literacy. “The biggest challenge is data,” says Iosad. “Getting the right data and setting up systems to use it effectively is key.” A weakness in the UK is the lack of integrated government datasets that can be linked to personalise services and support decision-making at all levels.
He points to a government blueprint that commits to giving every public sector dataset an API, enabling better data use and safeguarding privacy.
AI will continue to be adopted by local governments much like previous new technologies. While some challenges around training, supplier management, and contract oversight are familiar, transparency, scrutiny, and bias detection require fresh approaches. Local government must seek not quick fixes, but the best solutions to improve services and residents’ lives.
Your membership also unlocks:
