Local government data is not AI-ready by default, new report warns

Local authorities experimenting with artificial intelligence risk building on weak foundations, according to new research from the Open Data Institute (ODI) and technology company Nortal. Their joint study finds that while many councils are piloting AI to forecast demand, cut costs and improve services, most datasets remain unfit for algorithmic use.

The report, Insights from UK councils on standards, readiness and reform to modernise public data for AI, draws on interviews with council leaders, technical teams and programme partners, alongside analysis of public and operational datasets. It analyses ten local authority cases from Dorset to Leeds. It finds that councils making tangible progress share one trait: they approach data standards and infrastructure as strategic assets, not technical.

“Councils don’t need perfect data to make progress, but they do need the right kind of data for the job,” said Professor Elena Simperl, director of research at the ODI. “What works for search might not work for predictive modelling, and what works for predictive modelling won’t suit generative AI. Our framework moves the debate from asking whether councils are AI-ready to asking ready for what purpose, and what needs to change to get there.”

“AI-ready data is becoming the real infrastructure of modern government,” said Priit Liivak, Chief Government Technology Officer at Nortal. “Identifiers, metadata and versioned pipelines aren’t side issues—they’re what make services auditable, safe and capable of scaling.”

Three paths to readiness

The research identifies three complementary forms of readiness: search, machine learning and generative AI, each requiring different data conditions.

• Search readiness relies on structured, discoverable data with canonical identifiers such as UPRNs and clear metadata.
• Machine learning readiness depends on reproducible datasets, transparent lineage and bias documentation.
• Generative AI readiness requires context-rich corpora, segmentation and APIs that allow models to retrieve and reason across datasets.

The report notes that while many councils have improved discoverability through open data standards, far fewer have achieved the consistent, machine-readable infrastructure needed for predictive or generative systems.

A pattern of partial progress

Dorset Council’s use of structured social care data is highlighted as an example of predictive readiness, while London’s fire incident datasets show how consistent coding enables advanced analytics. Yet across most authorities, familiar barriers persist: inconsistent identifiers, missing metadata and limited API access. These shortcomings not only slow innovation but make it difficult to audit AI outputs or retrain models as circumstances change.

From pilots to policy

The report’s authors argue that councils should now move beyond pilots and invest in data quality, interoperability and governance as a shared foundation.

“The councils that standardise and document their data now will find that search, predictive and generative tools follow naturally,” said Liivak. “AI success starts with the data architecture, not the applications.”