A recent panel discussion brought together senior counter fraud professionals to examine whether government is effectively leveraging data analytics and automation to combat the £55-81 billion annual loss from fraud and error across the public sector.
The numbers are staggering. The GDS estimate that government could be making AI-driven savings of £45 billion annually by 2027; £6 billion of these are specifically from reducing fraud and error using AI and other data tools. And there are some fantastic use cases out there, but there is still a large gap between potential and practice.
Take the National Fraud Initiative. The system successfully checks your data against everyone else's and tells you what red flags to look at, so that you can investigate what is suspicious and whether any money can be recovered. But when the NAO audited over 400 central government bodies, only 36 actually use it. This is basic fraud prevention that we've suggested should be made mandatory across government.
We're seeing promising developments: the Department for Education now cross-references apprenticeship grants with tax records to verify employment, while DfT uses AI image detection to spot duplicate electric vehicle charging point claims. And a simple use case that more departments could implement is pay verification - only HMRC are using this at the moment to prevent mandate fraud, with DWP in the process of implementing it.
Sophisticated tools exist alongside very simple gaps. Government departments need to talk, and set up a coherent plan to capture these billions in potential savings.
Local authorities have embraced the mandatory NFI requirement, and we've built upon it. The London Fraud Hub has transformed the biennial NFI exercise into ongoing intelligence, particularly targeting dual working fraud where individuals hold multiple full-time positions.
Our collaboration with other London councils has expanded beyond payroll employees to include agency workers - previously a significant blind spot. Working with LBFIG and recruitment agencies, we're seeing strong detection rates and planning expanded matching capabilities.
The NFI isn't just about detection; it's prevention. When we identify fraud hotspots, we examine why those vulnerabilities exist and strengthen controls accordingly. The recommendation to make NFI mandatory across central government departments could significantly expand our collective fraud fighting capability.
We're witnessing growing expertise in data analytics across government organisations, though there's considerable room for expansion. GIAA is developing our own automation capabilities, particularly in trend and risk identification, to better serve the vast number of organisations we work with.
Our role involves connecting less developed organisations with those demonstrating best practice. The focus is on practical implementation - sharing what works and supporting capability building across the public sector landscape.
DWP uses well-tested data tools like the General Matching Service and real-time information systems that have proven their worth over time. But we're always looking for ways to improve and stay one step ahead of fraudsters.
We've developed a machine learning system that stops people from misusing universal credit advances before it happens. This year's Annual Report and Accounts shows our new eligibility verification system designed to tackle capital fraud and error - a major problem that needs advanced data analysis to solve. We're rolling out these machine learning models to more areas while continuing to use our existing, proven fraud prevention systems.
The public sector has strong data analytics tools and shows real success stories across different areas. Local government is working well together, and departments like DWP and HMRC are leading the way with innovative approaches. The building blocks are clearly in place.
The next step is joining up these efforts across government. This means making successful tools like NFI standard practice everywhere, rolling out basic protections like pay verification to all departments, and creating clear plans for using AI and data analytics more widely.
With £6 billion in potential savings from fraud prevention alone, government has both the tools and the financial incentive to make this work. The foundations are solid - now it's about scaling up what we know works.