Universal credit fraud: How AI could stop the crime – SAS

Following the news that millions of pounds of public money have been stolen in Universal Credit scams, comment below from Simon Dennis, director of AI & analytics innovation at data giant SAS, on the measures the government must take to stop the fraudsters in their tracks.



Simon Dennis, Director AI & Analytics Innovation, SAS UK


“Despite ambitions that Universal Credit (UC) would lead to a reduction in benefit fraud, today’s disappointing revelations indicate the opposite.  The success of these attacks casts a shadow over the Government’s anti-fraud strategy for UC and more specifically the counter -measures deployed by the Department for Work and Pensions (DWP).


“These scams appear to be far from sophisticated, but they are succeeding in significant numbers and the attacks described resemble similar attacks over a decade ago against the Tax Credits system, which led to that online system being shut down permanently.”


“Universal Credit has a degree of complexity that lends itself to a technology-based defence that can spot false claims at speed and at scale – often before the fraud has even been perpetrated – without laborious human intervention. Advanced analytics and AI-enabled anti-fraud systems can sift through huge numbers of claims, making connections with other government data such as validating landlord identity or spotting recurrences of certain typographical errors. It can then use these anomalies to identify potentially fraudulent activity in real time.


“The technology is readily available but it requires a ‘unified government’ approach to ensure that the right expertise, tools and data are assembled for the task. To illustrate this, consider the fact that some UC attacks would require collusion between individuals – such as between a claimant and their employer. In this scenario HMRC would have a view on the riskiness of the employer, yet if the DWP was acting in isolation it would have no way of knowing until its own systems recorded the crime – if indeed it were detected – and of course after the money had disappeared.


“Given that the volumes of data involved are immense, coming from a wide array of information sources, using AI to augment the human effort is a wise approach. Over time this AI capability will develop new techniques and this is one “headcount” that departments can share without cost. So the experience can be shared and, of course, the benefits will flow both ways.


“UC fraud is far from a victimless crime. People are being saddled with debts they cannot afford to repay, which again has echoes of tax credit overpayment debts of the past. It’s time that Government as a whole develops a united defence against UC fraud and other financial crimes, with AI providing the unifying core. Only then will over-stretched officials be able to focus on providing a better service for genuine claimants.”