Effective 1 January 2025, Vatican City State's Governorate issued its first AI rules — Decree n. DCCII — covering every civil department, from admin and health to police and courts, requiring labelled AI content and organisational-only judicial AI use.
Details
Promoter
Governorate of Vatican City State
Period
2025-
Keywords
public administration, AI governance, regulatory policy
Context
The Governorate of Vatican City State — the body running the sovereign territory's civil administration — sought to set ground rules for AI use across its workforce ahead of the EU AI Act's broader European rollout.
Activities
The Governorate issued Guidelines on Artificial Intelligence under Decree n. DCCII, effective 1 January 2025, applying across administration and HR, the Vatican Museums, gardening and refuse services, the fire brigade and gendarmerie, the post office, the health service and pharmacy, and the Vatican's court system, plus third-party suppliers and temporary contractors. Principles include that AI must serve, not replace, human beings; must not produce discriminatory or rights-violating automated decisions; and that AI-generated content must be labelled 'IA'. Judicial offices may use AI only for 'organisation and simplification' of research and casework, not for decisions.
Results
The guidelines direct the President of the Governorate's office to establish a five-person AI commission chaired by the Secretary-General, describing the document as a 'compass' ahead of further laws and implementing regulations still to come.
Conclusions
As a brand-new regulatory framework rather than a deployed AI service, there are no operational metrics yet, and independent legal commentary notes much operational detail is left to yet-to-be-published implementing regulations.
Implementation
Indicative cost
Low (< €50k)
Time to results
Short (< 1 year)
Staffing & skills
Five-person AI commission chaired by the Secretary-General of the Governorate
Conditions for success
Explicit carve-out restricting judicial AI use to 'organisation and simplification' rather than decisions
Mandatory 'IA' labelling requirement for AI-generated content to distinguish it from human work
Do you run this practice?
Claim it —
verified implementers get a public contact pathway and can propose corrections.
Data sources
Where this practice's information was retrieved from, and when.