Estonia's accidental €24 million ($27.4 million（約44億円）) gambling tax error in December prompted the rapid creation of Apsakaleidja, an AI tool that flags drafting mistakes in draft legislation. The success of the tool has led Prime Minister Kristen Michal to commit Estonia to broader integration of AI in lawmaking and government administration, with plans to potentially create official digital identities for AI agents operating in the public sector.
Summaries like this, in your inbox every morning.
Sign up free →What happened
In December, Estonia's parliament accidentally excluded online casinos from new gambling tax rules, costing the government €24 million ($27.4 million（約44億円）) in lost revenue for a year. Former undersecretary Luukas Ilves then built Apsakaleidja ("Fuckup Finder"), a tool that flags drafting errors—broken references, contradictions, arithmetic mistakes, faulty dates—in pending legislation; of 112 bills currently scanned, 102 are marked high risk.
Why it matters
The mistake revealed to Estonia's government that AI can reliably catch legislative oversights before passage. Prime Minister Kristen Michal has now committed the country to embedding AI tools in lawmaking and administrative automation, with the goal of doubling productivity by 2035. Estonia is already positioned for rapid AI adoption, since 99 percent of public services are already online and the country has invested heavily in digital identity infrastructure.
What to watch
In June, Michal announced that if current plans proceed, Estonia intends to become "the first country in the world to create official digital identities for AI agents." A bill granting state and local government the right to use AI for administrative automation is currently moving through parliament, with debate ongoing about which decisions (rule-bound vs. discretionary) should remain subject to human oversight.
Estonia's €24 million ($27.4 million（約44億円）) tax blunder in December—where a single misdrafted phrase accidentally removed online casinos from new gambling tax rules—became a catalyst for government AI adoption rather than a lasting scandal. The immediate, successful deployment of Apsakaleidja by Luukas Ilves demonstrated that large language models like Claude and Gemini could reliably identify legislative inconsistencies that human reviewers had missed. This success shifted the frame: rather than treating the error as an isolated failure, Prime Minister Michal framed it as proof that AI could strengthen democratic lawmaking by catching errors before passage, without replacing human authority.
Estonia's embrace of AI governance builds on decades of prior investment in digital infrastructure. The country has already moved 99 percent of public services online and built a sophisticated digital identity system—foundations that lower the friction for AI integration and reduce the need to retrofit legacy bureaucratic processes. The debate now centers not on whether to use AI, but on how to govern its use: which categories of decisions (rule-bound allocation of benefits vs. discretionary judgment about individuals) are safe for automation, and how to ensure accountability and the right to human review remain intact. This framing—embedding human oversight as a deliberate architectural choice rather than an afterthought—reflects awareness that the legitimacy of AI in government depends on visible chains of responsibility and the preservation of democratic control.
No discussion yet for this article
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
