The Netherlands plans to reduce its central administration by an estimated 25,000 full-time-equivalent positions in an AI-led government overhaul targeting €3.5 billion in annual savings from 2035. The Cabinet will publish specific AI productivity goals and savings plans in the first quarter of 2027.
The government announcement on 26 October sets a savings target of €900 million in 2027, rising over subsequent years. The estimated workforce reduction would be spread across several years, with police and defence among the areas exempted.
State Secretary for Kingdom Relations and Effective Government Eric van der Burg outlined the programme in a letter to Parliament. The Cabinet says both efficiency measures and strategic decisions about which activities government should undertake will be needed to meet the savings targets.
The programme includes scrapping or simplifying at least 500 roles annually, reducing the number of advisory bodies and policy evaluations, and making greater use of joint purchasing. Ministries will also standardise IT systems and reduce surplus office space.
For residents and businesses, the Cabinet proposes more integrated services and easier, secure data sharing between public bodies. It intends to develop a legal solution within two years to support proactive services, so information already held by government does not have to be submitted repeatedly.
These changes make the announcement broader than a plan to replace individual workers with AI systems. The estimated reduction is expressed in full-time equivalents, rather than a confirmed number of individual redundancies.
The overhaul follows a government procurement strategy, which places digital autonomy, security and continuity of IT services among purchasing priorities.
That strategy calls for risks to be identified before contracts are signed and monitored throughout delivery. It also proposes stronger contractual safeguards for digital autonomy and security. For AI suppliers, these priorities establish a procurement context that extends beyond productivity claims to operational resilience and technology dependence.
The relationship between deployment, accountability and ongoing monitoring also falls within AI governance. However, the October workforce announcement does not provide the detailed AI deployment programme; those targets and savings plans remain scheduled for early 2027.
Public-sector AI deployment also sits within the EU AI Act’s risk-based framework. The European Commission identifies employment tools and certain systems affecting access to essential public services as potentially high-risk uses. Its listed safeguards include risk assessment, documentation, human oversight, accuracy and cybersecurity. Public-sector use alone does not make every AI system high-risk.
Related standards address different aspects of organisational control: ISO/IEC 42001 covers AI management systems, while ISO/IEC 42005 provides an AI impact assessment framework. Neither is presented as a new certification requirement in the Dutch announcement.
The next milestones are the first-quarter 2027 AI plans and a multi-year civil-service programme due within six months of the announcement. The government says the detailed allocation of savings will follow in the spring budget update, providing a clearer account of how the proposed reductions will be implemented.