Jack FenwickPolitical correspondent
PA MediaA brand new synthetic intelligence software designed to crack down on fraud has helped the UK authorities get better virtually £500m during the last yr, the BBC can reveal.
Greater than a 3rd of the cash clawed again associated to fraudulent exercise in the course of the Covid-19 pandemic, with different money being recouped from illegal council tax claims and unlawful subletting of social housing.
The federal government will announce later {that a} new AI software which has helped to determine the fraud will now be licensed to different international locations, together with the US and Australia.
Civil liberties campaigners have beforehand criticised the Labour authorities for its use of AI in making an attempt to counter fraud.
The Cupboard Workplace says the £480m recovered within the 12 months from April 2024 is the most important sum ever reclaimed by authorities anti-fraud groups in a single yr.
The financial savings have been made by cross-referencing data held by completely different authorities departments, in addition to utilizing a brand new AI software.
Ministers say the financial savings will now be used to recruit nurses, lecturers and law enforcement officials.
Of the whole sum recovered, £186m was associated to Covid fraud.
Ministers have lengthy vowed to claw again a few of the cash misplaced in the course of the pandemic, however the £186m recouped is a fraction of the quantity Labour has beforehand mentioned was lacking.
Earlier than final yr’s basic election, the now-Chancellor Rachel Reeves claimed greater than £7bn of public cash was misplaced to fraud in the course of the pandemic.
These financial savings embody the blocking of lots of of hundreds of corporations with probably fraudulent Bounce Again Loans from dissolving.
Bounce Again Loans had been loans of as much as £50,000 arrange by the federal government in the course of the pandemic to assist companies.
However the scheme has been criticised for not being diligent sufficient and successfully inviting fraud, as corporations that dissolve earlier than paying again the cash typically then don’t must repay something.
The Cupboard Workplace says one case it discovered concerned a lady who invented an organization after which despatched the mortgage cash to Poland.
Anti-fraud summit
Cupboard Workplace minister Josh Simons will announce the financial savings at an anti-fraud summit held collectively by the UK, the US, Canada and Australia on Wednesday.
He mentioned “cutting-edge AI and knowledge instruments” would guarantee the federal government can defend public funds and never “line the pockets of scammers and swindlers”.
The brand new AI software, known as the Fraud Danger Evaluation Accelerator, was developed by researchers within the Cupboard Workplace and can now be rolled out throughout different authorities departments.
The Cupboard Workplace mentioned the software “scans new insurance policies and procedures for weaknesses earlier than they are often exploited” and claims it may make insurance policies “fraud-proof” earlier than they’re carried out.
It was developed due to concern inside Whitehall concerning the degree of fraud seen in the course of the pandemic.
Simons will announce that the UK authorities will now license the software for worldwide use, and it’s anticipated that the US, Canada, Australia and New Zealand will all undertake it indirectly.
However the transfer may trigger concern amongst marketing campaign teams already sad with the federal government’s use of AI.
Final yr, an AI software used to crack down on welfare fraud was discovered to point out bias in accordance with individuals’s age, incapacity, marital standing and nationality.
Paperwork launched to the Guardian beneath Freedom of Data Legal guidelines admitted the software utilized by the Division for Work and Pensions confirmed a “statistically vital consequence disparity” in a “equity evaluation”.
A report by Amnesty Worldwide launched earlier this yr criticised the federal government’s “unchecked use of tech and AI programs”.

