Warning UK ministers are putting interests of Big Tech ahead of public on AI
Ministers have been accused of favouring Big Tech firms over the public interest by spending billions on AI systems without adequate transparency, in a warning from MPs and rights groups including the Open Rights Group. Analysis by consultancy Tussell found the UK public sector has awarded more than 2,000 AI-related contracts since 2018, worth around £5bn, covering areas such as policing, housing, benefits and asylum decisions, yet campaigners say the public remains largely in the dark about how these systems affect them.
The Open Rights Group's new report highlights concerns including the DWP's use of machine learning to flag potentially fraudulent benefit claims, which the National Audit Office found disproportionately targeted older claimants and non-UK nationals, and a £330m NHS England contract with US firm Palantir to link patient data. It also notes Sadiq Khan's decision to block a £50m Met Police deal with Palantir in June over concerns about how the contract was won. The report calls for mandatory impact assessments and expert audits of government AI systems, while a government spokesperson said it was investing £100m to support British AI capability and had frameworks in place to ensure safe, accountable use with humans retaining control of final decisions.
- UK public sector awarded 2,000+ AI contracts since 2018, worth £5bn
- MPs and campaigners warn of opaque, unaccountable AI use affecting benefits, policing, asylum
- DWP fraud-flagging AI over-referred older and non-UK claimants, NAO found
New here? Start with this
Ministers, MPs and campaign groups such as the Open Rights Group are at odds over how the government uses artificial intelligence in public services. Research shows that since 2018 the UK public sector has signed over 2,000 AI-related contracts worth roughly £5bn, covering sensitive areas including policing, benefits, housing and asylum decisions, often involving large technology companies.
Concerns centre on a lack of transparency about how these systems work and affect people. Examples cited include the Department for Work and Pensions using machine learning to flag potentially fraudulent benefit claims, which official auditors found disproportionately affected older claimants and non-UK nationals, and a large NHS England data contract with the US firm Palantir.
The issue matters because such systems can influence decisions about people's benefits, healthcare, immigration status or policing, often without those affected knowing AI was involved. Campaigners are calling for stronger oversight, such as mandatory assessments of AI systems' impact, while the government maintains it has safeguards in place and that humans remain responsible for final decisions.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Campaigners, MPs and civil liberties groups argue that when AI systems help decide who gets benefits, housing, policing attention or asylum status, the public has a fundamental right to know how those decisions are made and to have them independently scrutinised. They point to the National Audit Office's finding that DWP fraud-detection algorithms disproportionately flagged older claimants and non-UK nationals as evidence that opaque systems can embed bias and cause real harm to vulnerable people without adequate recourse. Given the scale of spending, some £5bn across 2,000-plus contracts, and the sensitivity of the data involved, they contend that mandatory impact assessments and expert audits are a proportionate safeguard, not an obstacle, ensuring accountability keeps pace with rapidly expanding automated decision-making in public life.
The case against
Government officials and AI-adoption advocates argue that modern public services, from the NHS to policing, face acute pressure from ageing infrastructure, staff shortages and rising demand, and that responsibly deployed AI can improve efficiency and outcomes for millions of citizens. They note that frameworks already exist to govern procurement and use, that humans retain final decision-making authority rather than ceding it to algorithms, and that the £100m investment in British AI capability reflects a genuine effort to build sovereign expertise rather than simply outsourcing to Big Tech. From this perspective, heavy-handed mandatory audit requirements risk slowing beneficial innovation, deterring capable partners, and leaving Britain lagging behind other nations that are integrating AI into government more quickly, without necessarily making the public any safer.