🔍 Read the full analysis: AI And Public Administration In Europe: Jobs, Asylum And Online Responses on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
A Dutch government letter projects about 25,000 fewer central-government full-time jobs by 2035, but describes AI as one tool and warns against overstating its staffing effects. In Britain, a rights group says Home Office asylum summaries had higher error rates in a spot check than in a pilot; a UK consultation also found most submissions were bot-generated. Sweden has blocked public ChatGPT and Claude on government staff computers, while the scope and follow-up of the UK scrutiny remain unsettled.
A Dutch state secretary, Eric van der Burg, sent parliament a progress letter on October 26 describing plans to reduce central government running costs. The projected annual cuts rise from €901 million in 2027 to €3.49 billion from 2035. The letter calls the estimate of roughly 25,000 full-time positions a “rough estimate.” Police and defence are exempt, and the letter says the covered bodies employed 160,016 full-time equivalents at the end of 2025.
The Dutch letter lists AI alongside other measures, including simpler rules, shared desks, standardised IT and joint purchasing. It also cautions against expecting too much from digitalisation and AI, saying earlier automation did not deliver promised staffing and cost savings and brought new tasks and contractors. The government expects general efficiency measures to cover about a third of the planned reductions; the remainder requires decisions about which activities to stop. AI savings targets are due in the first quarter of 2027.
In Britain, the Home Office uses two generative AI tools in asylum casework: one summarises interview transcripts and another searches official country information. The Open Rights Group (ORG), a digital rights organisation, says both tools use OpenAI models through Microsoft’s Azure cloud. ORG reports that a Home Office spot check found 45 of 203 summaries failed an initial review and 30 failed a second review. The department’s 2025 pilot had recorded a 9% inaccuracy or incompleteness rate. ORG also says only about 203 summaries were checked against an estimated 70,000 processed from April through July.
A separate UK health consultation in 2023 received 118,756 replies. The health department classified 90,835—76%—as bot submissions and excluded them, according to the source report. Sweden’s Government Offices blocked browser access to public ChatGPT and Claude on staff computers from October 26; staff are directed to approved tools. The restriction concerns public services and data arrangements, not a general prohibition on AI.
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’s role in public decisions
These cases put different public interests at stake: government employment, the quality and transparency of asylum casework, and the integrity of public consultation. In asylum processes, errors in summaries may affect how a caseworker understands an applicant’s account even if a human remains responsible for the decision. Applicants’ reported lack of notice or access to correct summaries raises questions about how mistakes can be identified.
The consultation figures also show how automated submissions can affect the apparent weight of public feedback. Excluding the suspected bots changed the reported balance of responses, though the supplied material does not identify the consultation’s substantive result. Sweden’s restriction highlights a different policy concern: government offices may limit consumer AI services to manage data and contracting risks while allowing approved alternatives.
For readers, the distinction between an AI tool supporting work and an AI system determining outcomes matters. So does the difference between a budget-driven staffing plan that includes AI and job cuts caused specifically by automation. The available evidence does not establish that AI alone explains the Dutch reductions or that the British tools decide asylum claims.
Products Worth Considering
LG UltraGear 32GX850A-B 32-Inch 4K UHD Glossy OLED Gaming Monitor
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Three different government approaches
The Dutch plan is part of a broader drive described by the cabinet as a “decisive government.” The projected employment figure is tied to a rising spending reduction, not presented in the letter as a count of positions directly eliminated by AI. Its warning about past automation is a notable qualification to claims that technology can deliver the planned savings.
In Britain, the Home Office tools were introduced in stages: the policy-search tool reached decision-makers in July 2025, and the interview-summary tool reached caseworkers in April 2026, according to ORG’s account. The department says trained caseworkers make every decision. It also said the tools saved 23 minutes per interview transcript and 37 minutes per case during the pilot.
ORG published its report on October 12 after making parliamentary and freedom-of-information requests. The group says neither asylum tool appears in the UK’s algorithmic transparency register, despite government guidance calling for departments to list such systems. The EU AI Act treats AI used in asylum decisions and access to public services as high-risk; the supplied source says related duties are scheduled to start on December 2, 2027, following this year’s “digital omnibus.”
“Do not overestimate the staffing effects of digitalisation and AI.”
— The Dutch government’s progress letter to parliament
Questions over checks and disclosure
It is not clear whether the UK Home Office will expand its review of asylum summaries, notify applicants about the tools, or provide a way for them to correct summaries. The spot-check results cover 203 documents, and the source does not provide enough detail to determine how representative that sample is. ORG’s findings are attributed to the group; the Home Office’s account differs in emphasis by stressing human decision-making and reported time savings.
WIRED reported on October 27 that the Independent Chief Inspector of Borders and Immigration had asked ORG for more information while planning inspections for next year. The inspector did not confirm a probe, so the status and scope of any formal investigation remain unclear. The source also refers to a tribunal judge’s July comments but cuts off before describing them; no further detail can be confirmed here.
For the Dutch plan, the final distribution of job reductions across organisations and the amount attributable to AI have not been specified. Sweden’s restriction is confirmed for public ChatGPT and Claude on staff computers, but the source does not give further detail about approved tools or how the policy will be reviewed.
Inspections and savings plans ahead
The next stated Dutch milestone is the publication of AI targets and savings plans in the first quarter of 2027. Those plans may clarify what role the government expects AI to play, but the October letter already warns that efficiency gains alone will not cover the full spending reduction.
In Britain, the inspector’s planned inspections next year and any response from the Home Office may clarify oversight of the asylum tools. The source material does not provide a confirmed timetable for changes to applicant notification, access to summaries or the transparency register. The EU AI Act duties cited in the report are scheduled to apply from December 2, 2027.
Sweden’s Government Offices have already redirected staff toward approved tools, but no further review date is provided. Across the three countries, the next evidence to watch is whether governments publish measurable safeguards and results—not just stated efficiency aims—and whether those measures address errors, accountability and the use of public input.
Key Questions
Is the Netherlands cutting 25,000 jobs because of AI?
No. The government letter projects about 25,000 fewer full-time positions as a rough estimate linked to broad central-government cost reductions. It lists AI as one measure among several and warns against overstating its staffing effects.
Does AI decide asylum claims in Britain?
The Home Office says trained caseworkers make every decision and the tools do not decide claims. One tool summarises interview transcripts and another searches official country information.
What did the UK spot check find?
ORG says Home Office experts reviewed 203 summaries: 45 failed the first review and 30 failed a second review. The source compares this with a 9% inaccuracy or incompleteness rate in the 2025 pilot.
Why did Sweden block ChatGPT and Claude on staff computers?
The source says Sweden blocked the browser versions of the public services and directed staff to approved tools. It describes the policy as concerning contracts and data, not as a ban on AI generally.
When do the cited EU AI Act duties begin?
The source says duties for high-risk AI used in asylum decisions and access to public services are scheduled to start on December 2, 2027, after a change described as this year’s “digital omnibus.”
Source: ThorstenMeyerAI.com
