DWP turns to tech to tackle benefit fraud - but integration and fairness issues remain

The Department for Work and Pensions (DWP) is stepping up its use of data and machine learning in efforts to reduce benefit fraud and error, but outdated systems and fairness concerns are hindering progress, according to a new report from the National Audit Office (NAO).

The watchdog’s latest review found that while DWP has made “real progress” in reducing overpayments, the proportion of benefits paid incorrectly remains too high. The overpayment rate fell from 3.6% (£9.7 billion) in 2023-24 to 3.3% (£9.5 billion) in 2024-25, with Universal Credit errors dropping sharply from 12.4% to 9.7%.

Since 2020, government has committed £6.7 billion in funding for fraud and error activity, much of which has gone into modernising case reviews, hiring more counter-fraud staff and expanding data analytics capacity.

A key component of this digital shift is the use of machine learning models to flag potentially fraudulent claims for Universal Credit advances. The system, introduced in 2022, has been credited with saving £4.4 million and is reported to be three times more effective at identifying high-risk claims than manual sampling.

However, the NAO report notes growing concern around the fairness and transparency of these algorithms. A DWP fairness analysis published in July revealed that older claimants and non-UK nationals were being disproportionately referred for review. The department has pledged to refine its models and ensure proportionality in future updates.

Beyond fraud detection, the DWP’s wider Service Modernisation Programme aims to integrate data from multiple benefit systems to give staff a single view of claimant information. Currently, fragmented and legacy IT systems mean caseworkers cannot always access a full picture, limiting opportunities to prevent incorrect payments before they happen.

NAO head Gareth Davies said the department is “moving in the right direction” but urged DWP to finalise its fraud and error strategy and press ahead with plans for cross-government data standards. “With the increase in funding and focus on prevention, the next few years will be key to its success,” he said.