evidoria

← Back to browse

Good practice

Overheidsbrede Monitor Generatieve AI — the Netherlands' Government-Wide Tracker of GenAI Use by Civil Servants

Netherlands · The Hague · See the Netherlands profile

TNO's second annual census of generative-AI use across Dutch government found applications grew from 8 in 2024 to 81 in 2025 — 78% built for civil servants' own internal work, and 37 already in daily use, led by municipalities.

81
GenAI applications counted (December 2025)
8
GenAI applications counted a year earlier (2024)
34
Applications run by municipalities
6
Applications run by executive agencies (e.g. RVO)
5
Applications run by ministries
78 %
Share of applications serving internal staff work
17 %
Share of applications facing citizens directly
2 %
Share of applications serving businesses
37 (46%)
Applications already in daily operational use
Overheidsbrede Monitor Generatieve AI — the Netherlands' Government-Wide Tracker of GenAI Use by Civil Servants

Details

Maturity
Scaling
Promoter
TNO (Netherlands Organisation for Applied Scientific Research)
Period
2024–2025 (published Dec 2025)
Keywords
public administration, digital government, generative AI governance, workplace productivity

Context

Following the Dutch cabinet's 2024 government-wide vision on generative AI, the Ministry of the Interior and Kingdom Relations commissioned TNO to run a recurring census of every generative-AI application in use across national, provincial and municipal government. The monitor replaces anecdote with a shared evidence base for adoption decisions.

Results

The December 2025 edition counted 81 applications, up from 8 a year earlier — a near tenfold rise. Municipalities lead with 34 applications, followed by executive agencies such as RVO with 6 and ministries with 5. 78% of applications serve government employees' own internal work, 17% face citizens directly and 2% serve businesses, and 37 applications (46%) are already part of daily operational practice.

Conclusions

TNO flags that most of the counted applications run on US-built foundation models, and says future editions of the monitor will track the resulting questions of digital sovereignty, vendor lock-in and long-term control.

Implementation

Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful