The Albania Test: Why AI Cannot Fix a Government That Does Not Work
Artificial intelligence in public administration is an integrity multiplier, not an integrity substitute. Greece layered AI onto a functioning digital state and is reporting measurable gains. Albania layered an "AI minister" onto a procurement agency that prosecutors now accuse of running a kickback scheme, and the machine became a public relations shield. For military, intelligence, law enforcement, and policy leaders, the lesson is simple: before asking what AI can do for your government, ask what your government will do to AI.
This morning, CNN aired a Columbia University World Leaders Forum panel in which Fareed Zakaria put the question that follows every AI conversation to Greek Prime Minister Kyriakos Mitsotakis and World Bank President Ajay Banga: will artificial intelligence cause catastrophic job losses? The segment was titled "How Greece Is Using A.I. to Make Government Work Better." That is half the story. The other half is playing out in Tirana.
What Greece Actually Built
Greece did not start with AI. It started with plumbing. Over six years, the Ministry of Digital Governance consolidated public services onto a single portal, gov.gr, and built a national register of administrative procedures. Only then did it add intelligence on top.
The results are concrete. The gov.gr assistant, mAigov, launched in December 2023 and is trained on more than 1,600 services and over 3,270 administrative procedures, with support for 25 languages. The Independent Authority for Public Revenue uses AI to detect tax evasion. A 2024 law integrates AI into the Hellenic Cadastre to sort title deeds and reduce land registry delays. The OECD's "Governing with Artificial Intelligence" report lists Greece among its public sector AI frontrunners, with 19 active projects.
Nevertheless, the Greek prime minister is no cheerleader. Days before the Columbia panel, speaking to investors in California, he warned that job disruption will arrive faster than any society has planned for.
"No government and no society is prepared for the speed." Kyriakos Mitsotakis, Prime Minister of Greece
He also acknowledged that Greece operates inside European Union AI rules and cannot write its own code. The regulatory fight, he said, will be decided in the United States.
What Albania Actually Got
In September 2025, Albania introduced Diella, an AI avatar promoted from government chatbot to "Minister of State for Artificial Intelligence," with a mandate to make public tenders "100 percent free of corruption." One year later, the record deserves equal attention.
In August 2026, the Organized Crime and Corruption Reporting Project revealed that anti-corruption prosecutors suspect the leadership of AKSHI, the agency that built Diella, of rigging tenders in favor of a criminal group. Prosecutors allege kickbacks that included wine boxes filled with cash, and authorities ordered the seizure of more than 40 million euros in assets. No one has been indicted, the investigation is ongoing, and Diella's own development was not among the contested contracts. Those caveats matter.
Moreover, a year after the appointment, Diella's actual role remains undefined. The opposition has asked the constitutional court who is accountable when the system errs. The actress whose face and voice Diella borrows sued the government in April, saying she consented to a help desk, not a cabinet post. Meanwhile, with nearly all services pushed online, middlemen now charge citizens to navigate the portal, replacing the old bribe with a new fee.
"Who feeds the data, and who verifies its integrity?" Erjon Curraj, SCiDEV Center, Tirana
The Albania test is this: a government that cannot run an honest procurement office cannot buy its way to honesty with software.
Will AI Cause Catastrophic Job Losses?
The brutal answer is that no one knows, and anyone who claims certainty is selling something. However, the evidence as of this month is clearer than the headlines suggest.
No aggregate collapse yet. The Federal Reserve Bank of Dallas finds only a slight impact on overall unemployment so far. Even in the worst case for young workers, the effect amounts to roughly 0.1 percentage point since late 2022.
A real door is closing on the young. Stanford's August 2026 update of its "Canaries in the Coal Mine" study, built on payroll records, finds employment among workers aged 22 to 25 in highly AI-exposed occupations about 19 percent below where it would be had it kept pace with less-exposed peers. The decline comes from fewer people entering those jobs, not from layoffs.
The evidence is contested. Norwegian population-wide data shows no significant junior employment effect, and some researchers argue remote work explains the entry-level slump better than AI.
Announcements are rising. U.S. employers attributed more than 116,000 announced cuts to AI through August 2026, about 22 percent of all announced cuts, according to Challenger, Gray & Christmas.
Banga's answer to Zakaria's question carries a distinction worth remembering. His remit is emerging markets, where most people do manual rather than cognitive work. There, the World Bank's own research finds the threat of job disruption far less imminent. The opportunity, in his framing, is "small AI" that makes a farmer more productive.
Three Tests Before Any Government Deploys AI
For African and Nigerian policymakers watching Athens and Tirana, three questions should precede any procurement, tax, or identity system contract:
Is the underlying process already honest and documented? AI trained on corrupted records will automate corruption at machine speed.
Who is accountable when the system is wrong? If no official can be named, the deployment is theater.
Can an independent party audit the decisions? Albania's prime minister promised that nothing would be untraceable. A year later, no one can trace it.
An Analyst's View
The Columbia panel asked the jobs question because it is the question audiences want answered. The question practitioners must answer is narrower and harder. AI will not decide whether a state works. The state decides that. Technology only sets the speed at which the answer becomes visible. Greece spent six years earning the right to deploy intelligence. Albania spent one press conference. That gap is the entire agenda for sovereign AI capacity in the Global South.
The Bottom Line
The era of AI-enabled government has arrived. Whether it strengthens the state or merely dresses up its weaknesses depends on institutions, not algorithms. Leaders who build the plumbing first will see Greek results. Leaders who skip to the avatar will get the Albanian ones.
OSRS can help. Our team provides intelligence-driven assessments of AI governance readiness, procurement integrity, and digital public infrastructure for government, law enforcement, and private-sector leaders. Contact us to schedule a briefing or an AI-readiness assessment for your organization.
Enjoyed this article? Share it with a colleague who needs to see it. Stay informed by subscribing to our email list and following us on Google News, Twitter, and LinkedIn for more exclusive cybersecurity insights and expert analyses.
About the Author
Dr. Sunday Oludare Ogunlana is the Founder and CEO of OGUN Security Research and Strategic Consulting LLC, a Professor of Cybersecurity, a national security scholar, and a television commentator. He advises government, law enforcement, academic, and private-sector organizations on security strategy and the intersection of artificial intelligence and national security.
Intelligence. Protection. Strategy. www.ogunsecurity.com
Comments