A new blueprint for tackling discussions and decisions about how local authorities can safely and ethically navigate the rise of artificial intelligence has been launched by researchers at the University of Sheffield.
While AI technologies are often promoted as a transformational tool to streamline public services, local authorities face significant challenges in making well-informed decisions about AI and ensuring these systems are used responsibly. A new report published by the University of Sheffield’s Centre for Machine Intelligence, in partnership with Sheffield City Council, tackles exactly that.
The report, titled Scoping Responsible AI in Local Government, is the result of a collaborative workshop bringing together experts and officers from across South Yorkshire, including Barnsley, Rotherham, Sheffield, and the South Yorkshire Mayoral Combined Authority (SYMCA), alongside the Ada Lovelace Institute.
“For years councils have been under pressure to deliver more with less while also delivering transformational change" said Jonny Pearce, ICT Programme Manager at Sheffield City Council . "AI offers us real potential to modernise our ways of working; however, without clear, ethical frameworks and robust governance, that promise risks being undermined by unintended harm, erosion of trust, and inconsistent practice.”
Navigating market excitement
Authored by Denis Newman-Griffis, Shabnam Taghiyeva, and Susan Oman, the report advises councils to avoid adopting AI based purely on industry trends and market excitement. The researchers found that councils often struggle with a "one-size-fits-all" approach, sometimes rushing to buy expensive, proprietary AI software without a clear business case or a real understanding of how the tools work.
Furthermore, drawing on the University’s Public Voices in AI and FRAIM projects, the report highlights a gap in community trust. Residents, particularly those in high-stakes services like child protection or adult social care, are rarely consulted on data usage. Simon Gunnell, Head of Digital Access at Sheffield City Council warns that this can leave vulnerable individuals unable to understand or challenge automated decisions.
“For residents in high‑stakes services such as social care and child protection, decisions are not abstract, and have real-world impact" said Gunnell. "Any use of AI must ensure our service users are not excluded from understanding, questioning, and influencing how decisions are made.”
What’s in the report?
To help councils turn these challenges into opportunities, the report offers a practical, problem-led framework. Rather than buying a tool and looking for a way to use it, the authors recommend that councils first identify a specific challenge in their services and then evaluate whether AI is the appropriate fix.
To achieve this, the report recommends councils establish clear accountability and leadership structures, as oversight is often entirely missing. Because signing off on AI tools and managing risk will frequently involve multiple people, councils must formally define these roles. Additionally, the report highlights the need to train staff so they fully understand how AI tools work and how they will impact their daily roles.
Instead of every council trying to navigate this alone, the authors also suggest that neighbouring local authorities team up to share what they have learned and tackle similar regional problems together.
Most importantly, the framework stresses that councils must get their internal data organised and accurate before buying into any new technology, as automated systems are only as good as the information they rely on.
“AI is only as reliable as the data behind it; without clean, governed data, councils risk automating mistakes and amplifying them faster than ever before” said Simon Gunnell, Head of Digital Access at Sheffield City Council.
Finally, the authors argue that councils should steer away from vague promises of efficiency and instead set clear goals that demonstrate real-world benefits to residents. By bringing local people into the conversation from the start, councils can ensure that these new digital tools actively build community trust and transparency.
What’s next for South Yorkshire?
The initial workshop has already sparked the creation of the new South Yorkshire Partnership on Responsible AI.
Looking ahead, the team will be running a follow-up series of workshops this summer with all five South Yorkshire local authorities to put these recommendations into practice. With funding bids underway to develop new learning resources and scope out a larger national project, the region is rapidly positioning itself as a leader in public-sector AI governance.
“We’re excited to be working with Sheffield City Council and the wider local government community in South Yorkshire on shaping the future of responsible AI. South Yorkshire has an outstanding history of innovation, and we see the type of cross-council, cross-sector partnership we’re building through this work as a template for tackling the challenge of responsible AI in the public sector nationwide.
This report will help steer those discussions, and we look forward to building a national blueprint for the local government sector through the South Yorkshire Partnership on Responsible AI.” – Denis Newman-Griffis, Theme Lead for AI-Enabled Research, Centre for Machine Intelligence