Government

The sector that barely moved. Government institutions entered the AI era with strong editorial oversight, formal approval chains, and established style conventions. Those structures held. The result is the lowest drift of any sector in the corpus.

−3.2pts
Sector median shift from baseline
34%
Mean intra-sector similarity (was 28%)
8 organisations6,842 documents6 countries5 languages2015–2025

The one-minute story

Three findings from the Government sector corpus. Each carries statistical evidence from the full PRISM™ analysis.

−3.2pts

Smallest median drift. Formal editorial governance kept voice intact across all eight institutions in the corpus.

PRISM™ TEMPORAL · sector median drift
71.2

Cultural markers per 1,000 tokens. Highest of any sector. Dense with institutional references, legislative terminology, and national administrative language.

PRISM™ L1 · Cultural Markers
Post-2022

Where other sectors accelerated downward after ChatGPT, government barely changed. Approval chains and formal style guides acted as structural resistance.

PRISM™ TEMPORAL · year-over-year acceleration

Where Government organisations stand today

Current PLI scores versus 2015–2019 baseline. Sorted by drift magnitude.

PLI scores by organisation

Current PLI score versus 2015–2019 baseline. Scores with fewer than 10 documents in the current window are excluded. ⓘ Ordering rationale

OrganisationCurrent PLIBaseline PLICoverage
UK Gov78.180.2
Ministère76.479.8
Swiss Fed75.177.9
Bundesregierung74.877.5
Rijksoverheid73.976.4
Governo IT72.375.1
Gobierno ES71.674.2
EU Commission70.273.8

Organisations are ranked by PLI score. The confidence badge reflects the number of scored documents in the current window: High (≥ 30 documents), Medium (10–29), Low (fewer than 10). Low-confidence scores are included in the ranking and should be read alongside the document count.

Government PLI over time

Sector median with interquartile range (shaded). Corpus median shown as dashed reference line.

Which dimensions define stability

Ten PRISM™ dimensions. Baseline (dashed) versus current (solid). Sorted by drift in the table.

DimensionBaselineCurrentΔ
Cultural Markers82.071.0-11.0
Idiom Density70.066.0-4.0
Burstiness68.065.0-3.0
Lexical Diversity74.071.0-3.0
Perplexity72.070.0-2.0
Readability62.060.0-2.0
Jargon Load76.074.0-2.0
Citation Density71.069.0-2.0
Hedging Language65.068.0+3.0
Generic Phrases58.062.0+4.0

Government's linguistic neighbours

Cross-sector linguistic similarity. Higher scores mean Government is beginning to sound like other sectors.

Government Academic
38%
Cross-sector linguistic similarity. Was 24% in 2015–2019 (+14 points)
Government Healthcare
36%
Cross-sector linguistic similarity. Was 22% in 2015–2019 (+14 points)
Government Finance
32%
Cross-sector linguistic similarity. Was 18% in 2015–2019 (+14 points)
Government Industrial
28%
Cross-sector linguistic similarity. Was 16% in 2015–2019 (+12 points)

Corpus health for this sector

Quality metrics for the Government sector corpus used in this analysis.

6,842
Documents accepted
91%
Acceptance rate
2015–25
Year coverage
3
Archive sources
Sector data last updated February 2026 · Methodology version 2.1 · Full method · How to cite