Albania made history with Diella, the first AI minister overseeing procurement. From Brazil to Ukraine, nations are using AI to fight corruption and boost transparency.
In September 2025, Albania made history. For the first time, a country appointed an AI system as a cabinet-level minister. Meet Diella — the world’s first “AI minister” — who now oversees public procurement decisions. This bold move signals the beginning of a new era where artificial intelligence is not just a government tool but an active participant in policymaking.
But Albania is not alone. From Brazil’s ALICE system to Ukraine’s Dozorro, countries are experimenting with AI to fight corruption, cut red tape, and bring more transparency to public spending. The results are impressive — billions saved, audits that take days instead of months, and better detection of fraud. At the same time, these innovations raise big questions about accountability, ethics, and cybersecurity.
In this article, we’ll explore how AI is being used in government procurement, what Albania’s decision really means, and the lessons other nations can learn.
Download practical resources to make AI easier, faster, and more useful — starting now.
On September 11, 2025, Albania oPicially appointed Diella, an AI avatar originally developed for the government’s e-Albania portal, as Minister for Public Procurement. Diella was created by AKSHI (the National Agency for Information Society) using Microsoft’s Azure cloud infrastructure and advanced AI models.
Before her promotion, Diella had already processed over 36,600 digital documents and provided nearly 1,000 e-services to Albanian citizens. Now, she will evaluate tenders, flag risks, and publish procurement decisions — a role traditionally held by a human
minister.
This appointment has made Albania a global pioneer. Supporters argue that Diella will reduce political interference, make procurement decisions faster, and lower opportunities for corruption. Critics, however, question whether an AI system can hold legal responsibility or be held accountable if something goes wrong.
Albania may have the most headline-grabbing AI experiment, but other countries have already proven the value of algorithmic oversight.
In Brazil, the OPice of the Comptroller General (CGU) developed ALICE — the Bid, Contract and Notice Analyser. This AI tool performs preventive audits by scanning procurement data for irregularities.
The results are remarkable: ALICE can complete an audit in about eight days, compared to more than 400 days with traditional manual procedures. Between 2019 and 2022, ALICE flagged suspicious tenders leading to R$9.7 billion (≈USD 1.9B) worth of contracts being suspended or cancelled. That’s real money saved for taxpayers and a powerful deterrent for corrupt actors.
Ukraine’s fight against procurement fraud oPers another success story. The country’s open procurement platform, Prozorro, allows anyone to see public tenders in real time. On top of this, Dozorro, a civil-society-led monitoring system, uses machine learning to
flag risky tenders for human review.
Early tests of Dozorro showed a 26% increase in detecting unfair supplier selections and a 298% increase in collusion cases compared to manual reviews. This combination of AI analysis and citizen oversight is now seen as a model for participatory anti-corruption technology.
Not every country is ready to hand ministerial power to an AI, but many are investing heavily in AI infrastructure and governance:
These moves suggest that AI in governance is not just a trend but an emerging pillar of public administration.
For all their promise, AI systems like Diella face serious challenges that must be addressed if they are to succeed.
Who Is Responsible When AI Makes a Decision?
If an AI minister approves a contract that later turns out to be corrupt, who takes the blame? Most legal systems still require a human signature or final approval for binding decisions. This is why many experts argue that AI should assistministers, not replace them entirely.
Algorithmic Bias and Data Quality
If the data fed into AI systems is biased or incomplete, decisions could unfairly exclude legitimate bidders or fail to catch fraud. Governments need strict data governance, continuous monitoring, and “human-in-the-loop” review processes.
Cybersecurity Risks
Procurement data involves billions of dollars — a tempting target for hackers. AI models and training data must be secured against tampering. Governments should regularly “red team” their AI systems to find vulnerabilities before malicious actors do.
The experience from Albania, Brazil, Ukraine, and others points to clear best practices for using AI in procurement:
Publish decision rules, data sources, and audit trails so citizens can understand and trust AI decisions.
Keep humans legally accountable for final decisions, especially those with major financial or legal impact.
Track AI errors and maintain an incident database to catch systemic issues early.
Protect training data and models from attacks that could manipulate decisions.
Create clear mechanisms for suppliers and citizens to challenge or appeal AI-made decisions.
Follow global best practices, such as those in the EU AI Act, to harmonize standards.
AI has already proven its ability to save money, speed up audits, and improve detection of fraud. But Albania’s appointment of an AI minister is a bold experiment that tests how far society is willing to trust machines with public power.
Will other countries follow suit? Or will Diella remain a unique case study in techno- governance? Either way, the lesson is clear: AI in government is here to stay — and getting the balance right between innovation, accountability, and security will define the next decade of public sector reform.