Using AI to identify public sector fraud

When it comes to audits in the public sector, both accountability and transparency are essential. Not only is the public sector under increasing scrutiny to provide assurance that finances are being managed appropriately, but it is also vital to be able to give early warnings of financial pressures or failures. Right now, given the huge value of funds flowing from the public purse into the hands of individuals and companies due to COVID measures, renewed focus on audit is essential to ensure that these funds are used for the purposes intended by parliament.

As Rachel Kirkham, former Head of Data Analytics Research at the UK National Audit Office and now Director of AI Solutions at MindBridge, discusses, introducing AI to identify and rectify potential problems before they become an issue is a key way for public sector organisations and bodies to ensure public funds are being administered efficiently, effectively and economically.

Crime Wave

The National Crime Agency has warned repeatedly that criminals are seeking to capitalise on the Covid crisis and the latest warnings suggest that coronavirus-related fraud could end up costing the taxpayer £4bn. From the rise in company registrations associated with Bounce Back loan fraud, to job retention scheme (furlough) misuse, what plans are in place for government departments to identify the scale of fraud and error and then recoup lost funds?

There is no doubt that the speed with which these schemes were deployed, when the public sector was also dealing with a fundamental shift in service delivery, created both opportunities for fraud and risk of systematic error. But six months on, while the pandemic is still creating economic challenges, the peak of the financial crisis has passed. Ongoing financial support for businesses and individuals remains important and it is now essential to learn lessons in order to both target fraudulent activity and, critically, minimise the potential loss of public funds in the future.

Timing is everything. Government has an opportunity to review the last 6 months’ performance and strengthen internal controls to ensure that further use of public funds is appropriate. Technology should play a critical role in detecting and preventing future fraud and error.

Intelligence-Led Audit

If the public sector is to move beyond the current estimates of fraudulent activity and gain real insight into both the true level of fraud and the primary areas to address, an intelligent, data-led approach will be critical. The use of Artificial Intelligence (AI) in public sector IT systems can be used to detect errors, fraud or mismanagement of funds, and enable the process changes required to prevent further issues.

HMRC is leading the way, using its extensive experience in identifying and tackling tax fraud to address the misuse of furlough – an approach that has led to many companies making use of the amnesty to repay erroneous claims. Other public sector bodies, especially smaller local authorities, are less likely to have the skills or resources in place to undertake the required analysis. If public money is to be both recouped and safeguarded in the future, it is likely that a central government initiative will be required.

Data resources are key; the government holds a vast amount of data that could be used, although this will require cross-government collaboration and co-operation. It is possible that the delivery speed of COVID-19 responses will have led to data collection gaps – an issue that will need rapid exploration and resolution. It should be a priority to take stock of existing data holdings to identify any gaps and, at the same time, use Machine Learning to identify anomalies that could reveal either fraud or systematic error.

Taking Control

In addition to identifying fraud, this insight can also feed back into claims processes providing public sector bodies with a chance to move away from retrospective review towards the use of predictive analytics to improve control. With an understanding of the key indicators of fraud, the application process can automatically raise an alert when a claim looks unusual, minimising the risk of such claims being processed.

While many public sector bodies may still feel overwhelmed, it is essential to take these steps quickly. Even at a time of crisis, good processes are important – failing to learn from the mistakes of the past few months will simply compound the problem and lead to greater misuse of public funds. The public sector, businesses, and individuals need to learn how to operate in this environment, and that requires the right people to spend time looking at the data, identifying problems and putting in place new controls. With an AI-led approach, these individuals will learn lessons about what worked and what didn’t work in this unprecedented release of public funds. And they will gain invaluable insight into the identification of fraud – something that will provide on-going benefit for all public sector bodies.