Artificial IntelligenceBusinessNewswireTechnologyWhat's Buzzing

Estonia’s $28M AI Mistake That Created a “Fuckup Finder”

Originally published on: July 9, 2026
▼ Summary

– A drafting error in Estonia’s Gambling Tax Act accidentally excluded online casinos from taxation for a year, costing the government €24 million in revenue.
– The mistake was quickly spotted by AI tools (Claude and Gemini), prompting former official Luukas Ilves to build a prototype called “Fuckup Finder” that flags legislative errors.
– The incident led Estonia’s Prime Minister to advocate for AI use in government, launching the Eesti.ai program to double productivity by 2035.
– Estonia is pursuing legislation to allow AI agents to automate administrative processes and aims to create official digital identities for AI agents.
– The error raises questions about why human review failed during the drafting process, highlighting a need for better oversight in legislation.

In December, a single drafting error in Estonia’s Gambling Tax Act cost the government an estimated €24 million ($27.4 million) in lost revenue. The amendment, passed by the Riigikogu (Estonia’s parliament), was intended to lower the tax rate on remote gambling. However, the text mistakenly referenced only “skill games” for that year, entirely omitting games of chance and remote gambling. This oversight accidentally exempted online casinos from taxation for a full year.

The mistake was first caught by a legal counsel working for a gambling operator. But the real embarrassment came when Luukas Ilves, Estonia’s former undersecretary for digital transformation, decided to test the legislation using Claude and Gemini, two popular AI models. Both systems instantly spotted the inconsistency.

Within hours, Ilves developed a prototype tool he named Apsakaleidja, or “Fuckup Finder.” The tool pulls draft bills from the Riigikogu website and flags issues like broken references, contradictory phrasing, arithmetic errors, and impossible dates. It categorizes problems as high, medium, or low risk. Of the 112 bills currently listed on the platform, a staggering 102 are rated high risk. One example highlights contradictory wording within a draft text. Ilves even demonstrated the tool live on national television, leaving the host visibly surprised.

The blunder was embarrassing, but it also sparked a major shift inside the government. “The situation demonstrated that AI can be an incredibly useful assistant,” said Kristen Michal, Estonia’s prime minister. “And,in the form of a vibe-coded platform to check draft legislation created in response to the incident,we saw an example of how agentic tools can empower civil society and individual citizens.”

Estonia quickly doubled down on AI adoption. In January, Michal suggested the country might use tools like Apsakaleidja to draft legislation proactively, catching and fixing loopholes before laws are enacted. He launched the Eesti.ai program, designed to upskill Estonians in AI use, with the ambitious goal of doubling national productivity by 2035. Among the program’s advisers are Bolt founder Markus Villig and Ilves himself, the creator of the Fuckup Finder.

By April, the government submitted a bill to parliament granting state and local governments the right to use digital solutions,including AI,to automate administrative processes. That bill is currently under debate, with the intention of becoming law. In June, Michal told an Eesti.ai meeting that, if all goes according to plan, “Estonia will become the first country in the world to create official digital identities for AI agents.”

“This is a new environment for the public sector,” Michal explained. “It demands agility and the ability to adapt as technology changes.” Estonia is uniquely positioned to handle this transition. It has long led the way in digital identity integration, with 99 percent of public services already available online. As WIRED first reported a decade ago, Estonia is widely regarded as a model for the modern digital state. That foundation now makes AI adoption smoother. “Those investments now allow us to move faster and more confidently into the AI era,” Michal said.

Not everyone is fully convinced. Catherine Flick, a researcher in technology ethics at the University of Staffordshire, raises a fundamental question: “Why are humans not doing this review process as part of the legislation drafting procedure?” she asks. “At some point someone has to sit down and read through the whole thing, with the understanding of the context and all that sort of stuff, in order to make sure that this is a decent law.”

(Source: Wired)

Topics

ai in government 95% legislative errors 92% ai error detection 90% estonian digital government 88% gambling tax policy 85% tax revenue loss 83% ai agent identities 82% productivity goals 80% vibe coding 78% ai adoption challenges 77%