The UK government has halted its practice of automatically suspending benefit claims which are flagged by its AI-powered fraud detector, according to a senior official from the Department of Work and Pensions (DWP). The technology is used to identify claims to Universal Credit which may be fraudulent. Previously, the appplicant’s claim would be put on hold while officials investigated further. However, this is no longer the case, according to Neil Couling, a senior DWP civil servant, who revealed the change in approach in a hearing before a committee of MPs.
The DWP has placed an AI tool at the centre of its plan to counter growing levels of benefit fraud. The tool uses machine learning to analyse historic data, meaning it can identify higher-risk claims, which are then referred to officials for investigation. The DWP estimates that using AI could reduce benefit fraud by £1.6bn by 2030/31. It has been sought for screening claims for UC advances since 2022 and is continuing to be rolled out to help detect fraud in other areas, such as incorrect earnings and housing cost declarations.
Campaigners have cautioned that some level of bias in the tool may lead to delays in payments for legitimate claims. The DWP has never disclosed the number of claims which have been flagged by the system, or how many suspensions have been applied as a result, making its impact difficult to assess. Child Poverty Action Group has previously raised concerns and believes that MPs require reassurances that the automatic suspension of claims will not recommence.
The DWP said that it has processes in place to detect potential bias and that it carries out checks as effectively as possible. Although it has not publicly disclosed exactly how the algorithm works, it has pledged to publish more information about its impact in its next set of annual accounts, which are due later this year
Read the full article from The BBC here: Read More