The government has given the DWP £6.7bn in funding to tackle fraud and error between 2020 and 2029.
Since 2022, the DWP has mostly used this funding to increase its counter-fraud staff, expand its use of technology, and scale up its Targeted Case Review (TCR) programme to detect and correct fraud and error in existing universal credit claims.
More than one million claims have been reviewed through this programme, leading to savings of more than £581 million by March 2025. This exceeded DWP’s savings expectation by 11%.
The total that DWP expects to save from the TCR has increased over time – from an initial target of £2bn in savings by 2026-27, to £13.6bn by March 2030.
It is not only saving money by catching out fraud but also errors made by the DWP in overpaying benefit claimants, which can have a devastating impact on individuals who are forced to pay back the money they have been given, as the Big Issue previously reported.
Around £9.5bn in benefits was overpaid due to both fraud and error in 2024/2025, but that is just 3.6% of total payments – and down from 3.6% in the previous year. The estimated universal credit overpayment rate dropped from 12.4% in 2023/2024 to 9.7% in 2024/2025.
Advertising helps fund Big Issue’s mission to end poverty
Gareth Davies, head of the NAO, said: “The DWP has made real progress in tackling the levels of benefit overpayments due to fraud and error, but there is still a way to go. With the increase in funding and the greater focus on prevention, the next few years will be key to its success in addressing this long-standing issue.”
The government’s Public Authorities (Fraud, Error and Recovery) Bill, which details its plans to tackle fraud and error, is set to face a final debate in the House of Lords before amendments are considered and it reaches Royal Assent, after which it will be made law.
NAO is calling on the DWP to “finalise its approach to implementing its fraud and error strategy and progress its ambition to reduce the overpayment rate to the pre-pandemic level”. It also urges the department to build on its existing use of data analytics to explore how these emerging technologies may help to detect and prevent fraud and error.
Minister for transformation Andrew Western said: “We are proud of the progress we have made in reducing fraud and error in the benefits system. The overall rates have dropped for the first time in two years, and we will go further with our Fraud Bill, which is part of wider plans that will save £9.6bn by 2030.
“As the head of the NAO has recognised, we’re rightly making use of technology, and our one machine learning model has made us significantly more effective. There are numerous safeguards in place, and final decisions in relation to fraud and error are always made by a human.”
Yet there are aspects of the bill which have caused significant concern. In particular, there are fears over government plans to give the DWP greater powers to request information from benefit claimants’ bank accounts and deduct funds directly from them, as the Big Issue has extensively reported.
Advertising helps fund Big Issue’s mission to end poverty
Ayla Ozmen, director of policy and engagement at anti-poverty charity Z2K, said: “While we support efforts to drive down fraud and error in the benefits system, we have serious concerns about some of the measures the government is introducing as part of its plan to tackle this.
“Given that DWP decision-making when it comes to suspected overpayments is often poor, we fear that proposals to require banks to share information about claimants’ accounts could lead to people in clear financial need having their payments wrongly stopped. We are calling for the government to introduce proper safeguards to ensure no one faces the risk of being pushed into destitution by this measure.”
A disabled woman who spoke to the Big Issue was accused of owing the DWP £28,000, with officials alleging that she had been overpaid her disability benefits. This was proven to be a mistake and Big Issue’s story led to the debt being overturned and her benefits reinstated.
Rick Burgess, of the Greater Manchester Coalition of Disabled People, said: “The conflation of error and fraud into one figure is a long term problem the DWP seems unwilling to resolve. It also serves a political function to provide a narrative for further surveillance, harsher rules and punishments, and cuts to social security.
“We need a supportive social security system, instead what we have got is a shadow penal system for people in poverty. It is being normalised that people receiving DWP awards have fewer human rights than the rest of the population, our bank accounts will be spied on and we lose the right to privacy.
“The pathological prioritising of fraud narratives is part of justifying abuse by the DWP, it is a distraction from the inadequacy of benefits and the continuing cover up of deaths caused by the system.”
Advertising helps fund Big Issue’s mission to end poverty
A coalition of human rights, disability rights and older people’s advocacy groups have been campaigning against the government’s plans to tackle fraud and error within the welfare system, led by Big Brother Watch. Last week, it wrote a joint letter to ministers, warning that the reliance on algorithmic fraud detection could lead to “innocent” people being “wrongfully” accused.
The DWP’s own data has found bias in its machine learning model. It showed that older claimants (in age groups 45 to 54 and above) and non-UK nationals were being over-referred for review, with these groups being more likely to be asked to find and provide additional evidence for their claim.
Jake Hurfurt, head of research and investigations at Big Brother Watch, said: “Another push for an expansion of DWP algorithms at a time when the department’s internal assessments have found existing tools to be riddled with bias, is alarming. DWP algorithms are hidden behind a wall of secrecy and the people they affect know almost nothing about how they are profiled, and how data analytics influence decisions about them.
“Instead of pressing forward the DWP should take a step back and focus on fixing the biases in its existing tools, and become much more transparent about the potential harms to data rights posed by the mass-scale use of AI. It is wrong to subject millions of innocent people to shadowy automated or algorithmic decisions, and refuse to explain how these work.”
In terms of performance, DWP found the machine learning model to be around three times more effective at identifying fraud risk than a randomised control group sample. It concluded that it therefore “remains reasonable and proportionate to continue running the model”.
Amnesty research from this year explored how benefit claimants often feel like they are being “treated with suspicion” – with stories ranging from a claimant forced to attend an appointment just two days after their baby had died, to a deaf man’s disability benefits claims being terminated for not being able to attend a telephone appointment.
Advertising helps fund Big Issue’s mission to end poverty
Clark argues that automation and AI are “making this significantly worse, with little oversight and devastating consequences”.
“People are wrongly flagged for fraud, sanctioned due to system glitches, and pushed into financial crisis by flawed algorithms,” Clark said. “Needing to rely on social security is something that could happen to any of us and that we all have a right to. If the government needs to raise money it should prioritise taxing those with the broadest shoulders, rather than trampling over the basic rights of those who are in need.”
Do you have a story to tell or opinions to share about this? Get in touch and tell us more.
Reader-funded since 1991 – Big Issue brings you trustworthy journalism that drives real change.
Every day, our journalists dig deeper, speaking up for those society overlooks.
Could you help us keep doing this vital work? Support our journalism from £5 a month.
Advertising helps fund Big Issue’s mission to end poverty
