Question to the Department for Work and Pensions:

To ask the Secretary of State for Work and Pensions, how many claims were referred for investigation by AI or machine-learning models in each year since 2023; what proportion were upheld; and what the referral and outcome rates were by protected characteristic.

The Department does not use artificial intelligence or machine learning models to refer claims directly for fraud investigation.

The Department has strong controls in place to tackle fraud and deliver savings. Since the Autumn Budget 2024 we have committed to delivering gross savings of £14.6bn up to the end of 2030 – 31 from fraud, error and debt activity.

Our activities include:

• investments to deploy up to 3,000 additional staff within counter fraud teams,

• the extension of Targeted Case Reviews (TCR) in Universal Credit to check accuracy of claims at risk of being incorrect,

• implementation of Pension Credit Claim Reviews (PCCR) to check claims at risk of being incorrect,

• the introduction of the Public Authorities (Fraud, Error and Recovery) Act 2025 (PAFER) which introduces new powers to better identify, prevent and deter fraud and error

• and continuing to seek prosecutions in the courts for those who commit benefit fraud.

Overall, levels of benefit fraud are the lowest since the pandemic and currently stand at 2.2%, compared to a peak of 3.0% in FYE 2022. This demonstrates that the Department has strong controls in place to tackle fraud and deliver savings.