🔍 Read the full analysis: Europe’s Governments Open The Door To AI In Public Administration on ThorstenMeyerAI.com
Get the little things that make your day delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Developments reported across the Netherlands, Britain and Sweden show governments taking different approaches to AI in public administration. They include Dutch plans to reduce central-government staffing, UK use of AI tools in asylum casework and a Swedish block on public ChatGPT and Claude accounts for government staff. The source reports concerns about accuracy and public consultation responses, but does not establish a single Europe-wide policy or outcome.
Recent decisions in the Netherlands, Britain and Sweden show European governments adopting, examining and restricting artificial intelligence in public administration, according to a report published October 29. The developments range from Dutch plans to cut central-government costs and staffing to UK use of AI in asylum casework and Sweden’s block on public ChatGPT and Claude accounts for government staff.
In a letter to parliament dated October 26, Dutch State Secretary Eric van der Burg set out plans to reduce central-government running costs, reaching €3.49 billion a year from 2035. The letter gives a rough estimate of around 25,000 fewer full-time positions. It lists AI alongside other possible efficiency measures, including simpler rules, standardised IT and joint purchasing; it does not describe the staffing reduction as an AI-only cut.
The Dutch letter also warns against expecting too much from digitalisation and AI. It says earlier rounds of automation did not deliver promised savings in staff and costs, and that new tasks and contractors followed. The cabinet plans to publish AI targets and savings plans in the first quarter of 2027. The source says the staffing figures concern covered ministries, inspectorates and agencies; police and defence are exempt.
In Britain, the Home Office uses two tools in asylum casework: one summarises interview transcripts, while another searches official country information. The Open Rights Group (ORG), a digital rights organisation, says documents it obtained show the tools use OpenAI models through Microsoft’s Azure cloud. The Home Office says trained caseworkers make decisions and the tools do not decide asylum claims. Applicants are not specifically told AI was used, according to a statement by Minister Anna Turley cited in the report.
Infographic 01 · The Netherlands
25,000 fewer civil servants. AI is one tool, not the reason.
The driver is a budget cut. The government plans to use AI, and warns against expecting too much from it.
The cut grows every year
Cuts to the running costs of central government, € billion per year.
Police and defence are exempt. Many organisations face about 20% fewer staff, some up to 45%.
Where the money must come from
The letter’s own split of the cut.
AI is one of several tools inside both parts, next to simpler rules, standard IT, shared desks and joint purchasing.
AI in the plan
- “Ambitious targets and savings plans” for AI in Q1 2027
- Purchasing to examine “systematically” where AI fits
- Conditions: security, data quality, sovereignty
Also in the plan
- At least 500 rules a year scrapped or simplified
- One desk per two employees
- Fewer advisory bodies (28 councils, ~150 committees)
The letter’s own warning
- Past digitisation did not deliver the promised staff cuts
- It brought new tasks and more contractors
- Cuts may affect service and continuity
“Do not overestimate the staffing effects of digitalisation and AI.”Lesson 2 in the Dutch government’s own letter, 26 Oct 2026 · translation: Thorsten Meyer
Sources: BZK, Voortgangsbrief Slagvaardige Overheid + annex, 26 Oct 2026; Rijksoverheid news release, 26 Oct 2026. 25,000 ÷ 160,016 ≈ 1 in 6 and bar scaling: Thorsten Meyer.
Powered by Thorsten Meyer AI
AI Enters High-Stakes Government Work
These cases show that government AI is not limited to administrative experiments. It can affect how officials handle public money, asylum claims and public consultation—areas where errors or unreliable inputs can have consequences for people and policy.
The British case raises questions about oversight because the summarisation tool is used to condense interviews that may inform asylum decisions. ORG says its review found more errors in a later spot check than in the pilot, while the Home Office maintains that people make the decisions. The difference matters: assistance with a case file is not the same as an automated decision, but inaccurate summaries could still influence how a caseworker understands an account.
The Dutch plan also illustrates why job-cut announcements need careful attribution. The government’s target is tied to a broad budget reduction, with AI named as one potential lever. The same letter cautions that automation may not produce the staffing savings expected. Sweden’s restriction, meanwhile, shows a government setting limits on which consumer AI services staff can access, rather than abandoning AI use altogether.
Products Worth Considering
Philips Hue Secure Wired Video Doorbell, 2K, Black
Dreame L50 Ultra Robot Vacuum and Mop
Bose S1 Pro+ Portable PA Speaker
As an affiliate, we earn on qualifying purchases.
Three Governments, Different Controls
The developments described are not a single coordinated European policy. In the Netherlands, the October 26 letter sets out a long-term spending and workforce plan. Its estimated 25,000 positions are a rough translation of the budget cuts, not a stated count of jobs that AI itself will eliminate. The source reports that covered organisations employed 160,016 full-time-equivalent staff at the end of 2025.
In Britain, the asylum summarisation tool was made available to all caseworkers in April 2026, while the policy-search tool reached all decision-makers in July 2025, according to the report. ORG published its findings on October 12 after parliamentary questions and freedom-of-information requests. It says the 2025 pilot found 9% of summaries inaccurate or incomplete. In a 2026 review, Home Office experts initially marked 45 of 203 summaries as failures, or 22%; a second review found 30 failures, or 15%.
The source also says a 2023 UK health consultation received 118,756 responses, of which the department flagged 90,835—76%—as bots and excluded them. It reports that removing those responses changed the consultation’s result, but provides no detail here on the policy question or the method used to classify responses. Sweden’s Government Offices blocked browser access to public versions of ChatGPT and Claude from October 26, directing staff to approved tools instead. The report characterises this as a contract and data-control measure, not a general ban on AI.
Accuracy, Oversight and Impact
Several questions remain unresolved in the material available. The Home Office’s reported sample of 203 summaries is small relative to the tool’s reported monthly use of about 17,515 cases; the source estimates that roughly 0.3% of summaries were checked. It is not clear whether the spot check represents error rates across all cases, how errors were defined, or whether any affected applicants experienced a change in their cases.
The report says the tools are not listed in the UK’s algorithmic transparency register, despite guidance asking departments to disclose such systems. It also says the Independent Chief Inspector of Borders and Immigration asked ORG for more information while planning inspections for next year. That does not confirm that the inspector has opened a probe. The source’s final account of a tribunal judge’s comments is incomplete, so no further conclusion about the ruling can be drawn here.
For the Netherlands, the projected job reduction is a rough estimate linked to a budget plan, and the eventual contribution of AI is not yet specified. For the UK consultation, the report does not provide enough information to independently evaluate how responses were identified as bots or the policy effect of excluding them. The longer-term results of Sweden’s restriction are also unknown.
Plans and Reviews Still Ahead
The Dutch cabinet is expected to publish its AI targets and savings plans in the first quarter of 2027. Those plans should clarify which government functions could use AI and how the cabinet expects the technology to contribute to savings. The broader cost reductions extend through 2035, leaving substantial time for the estimates and policies to change.
In Britain, the Independent Chief Inspector of Borders and Immigration is planning inspections for next year and has requested further information from ORG, according to the report. Whether those inspections will examine the asylum tools, and what any review may find, has not been confirmed. The Home Office’s position is that trained staff remain responsible for decisions.
Sweden’s Government Offices have directed staff toward approved services following the block on public ChatGPT and Claude. The report does not give a date for reviewing that restriction or details of the approved alternatives. Across all three countries, the next developments to watch are the publication of Dutch implementation plans, any UK inspection findings, and further information on how governments measure accuracy, disclose AI use and protect public data.
Key Questions
Is the Netherlands cutting 25,000 jobs because of AI?
Not according to the government letter described in the source. It gives a rough estimate of around 25,000 fewer full-time positions as the budget cuts grow, and lists AI as one of several possible efficiency tools. The letter warns against overestimating AI’s staffing effects.
Does AI decide asylum claims in Britain?
The Home Office says trained caseworkers make every decision and that its tools do not decide claims. One tool summarises interview transcripts and another searches country information. ORG has raised concerns about accuracy, transparency and the limited number of summaries checked.
How many asylum summaries failed the UK spot check?
Home Office experts reviewed 203 summaries. According to the source, 45 failed the first review, or 22%; a second review found 30 failures, or 15%. These figures come from a spot check and do not establish the error rate for every case.
Has Sweden banned AI in government?
No general ban is described. The Government Offices blocked browser versions of ChatGPT and Claude on staff computers from October 26 and directed employees to approved tools. The source says the measure concerns contracts and data handling.
What happens next in these cases?
The Netherlands plans to publish AI targets and savings plans in the first quarter of 2027. A UK border-inspection body is planning inspections for next year, but a formal probe into the asylum tools is not confirmed. The source gives no review date for Sweden’s restriction.
Source: ThorstenMeyerAI.com
