France is accelerating the digitalization of public administration — automation of procedures, growing use of AI in administrative decision-making, and ongoing debates about algorithmic governance.
But in Tirana, Albania just crossed a line no other state had before: on September 11, 2025, the Albanian government appointed an artificial intelligence named “Diella” as Minister of Public Procurement.
⚙️ An unprecedented event
“Diella” — meaning sun in Albanian — was originally a virtual assistant on the e-Albania platform.
She is now tasked with evaluating and awarding public contracts, with an official mission: “ensuring 100% corruption-free management.”
A world first.
Yet behind this innovation lies a fundamental question: can a sovereign function be entrusted to an entity without legal personality or political responsibility?
⚖️ Constitutional implications
The principle of ministerial responsibility requires that a member of government be a natural person — accountable to Parliament and liable for political or even criminal sanctions.
An AI, lacking legal personality, can neither take an oath nor be held to account.
This appointment therefore collides head-on with the very foundations of Albanian constitutional law and, more broadly, with European democratic standards.
France’s Conseil d’État, in its 2022 report “Artificial Intelligence and Public Action”, stressed that the use of AI must always operate within a framework of human accountability and algorithmic loyalty.
A virtual minister, by its very nature, contradicts that principle.
🏛️ Public procurement law challenges
Awarding a public contract constitutes an administrative act producing legal effects on third parties.
Under Article 22 of the GDPR, no one shall be subject to a decision based solely on automated processing that produces legal effects concerning them, unless appropriate safeguards are in place.
Such safeguards require effective human oversight.
Without a clearly defined framework for Diella’s intervention — supervision, appeal procedures, traceability — any procurement award could be deemed unlawful.
Excluded bidders would have a strong legal basis to challenge such decisions for lack of competence by the issuing authority.
🤖 Algorithmic bias and regulatory risks
Since its adoption in 2024, the European Artificial Intelligence Act (AI Act) classifies AI systems used in the allocation of public services or benefits as high-risk systems.
These require decision traceability, algorithmic explainability, and continuous human oversight.
No public information confirms that Diella meets these conditions.
This creates a double vulnerability:
- Potential bias or discriminatory errors in contract evaluation;
- Legal fragility of decisions made by a system without any clear line of responsibility.
🔭 A European paradox
While Albania seeks to prove its modernity on its path toward EU accession, this appointment could in fact undermine its alignment with the EU acquis communautaire:
absence of legal personality, possible non-compliance with the GDPR, and potential infringement of the right to an effective remedy guaranteed by Article 47 of the Charter of Fundamental Rights of the European Union.
It perfectly illustrates the delicate balance European states must strike — between technological ambition and the preservation of the rule of law.
💬 The Albanian experiment highlights the urgent need for a clear legal framework governing AI in sovereign functions.
Appointing an AI to ministerial office, without legal personality or accountability, effectively undermines the principle of legality and blurs the democratic chain of responsibility.
The challenge is no longer technological, it is constitutional.
