Estonia to Become First Country to Give AI Agents Personal ID Codes

Estonia is preparing to roll out something no other country in the world yet has: separate identification codes for artificial intelligence agents. Prime Minister Kristen Michal announced on June 17, 2026 that autonomous AI systems acting on behalf of citizens and companies will receive their own verifiable digital identity, separate from the identity of their owners.

This addresses a situation that's becoming increasingly common: more and more users and companies are letting large language model based programs fill out tax filings, prepare reports, or communicate with government and banking IT systems on their own. The problem is that today such an agent has no legal identity of its own. It has to borrow its owner's digital identity, which in practice means granting it full access to accounts, services, and data.

How the AI code would work

Rather than handing an agent the keys to everything, an owner would assign it a narrow set of permissions tied to their verified identity. The identification code would precisely define what a given agent is allowed to do: whether it may only view a document, draft one, or authorize a payment within a preset spending limit.

In the future, artificial intelligence will increasingly carry out digital tasks for us, drawing up reports, preparing tax filings, or communicating with information systems - Kristen Michal, Prime Minister of Estonia

Michal also highlighted the core problem the new system is meant to solve. As he put it in comments cited by foreign media, no one should be forced to give their AI assistant access to all of their permissions. Instead, the mechanism would leave a trail that can be checked after the fact, when it's necessary to establish who is actually responsible for a given action.

Estonia's digital foundation

Estonia isn't starting from scratch. For two decades the country has been building one of the world's most advanced digital government systems, based on personal identification codes for citizens and the X-Road data exchange platform, which links public registries and private companies. Estonia also runs the e-Residency program, letting foreigners run digital companies remotely via a mobile ID, has deployed AI chatbots built with OpenAI in every school, and operates the Bürokratt network of agents that handles public services.

That foundation is meant to make rolling out the new system faster than in countries still building basic digital identity infrastructure from the ground up. The government is also hoping the solution becomes a reference point for other countries, much as Estonia's e-Residency model did before it.

Expert reaction

The digital identity verification industry is watching Estonia's plan as a precedent that could set the standard for the entire sector. Philipp Pointner of Jumio said Estonia's decision to give AI agents identification numbers sets a template for auditing what such agents can actually do.

Estonia's decision to assign digital identification numbers to AI agents is setting a precedent on how we must audit what agents can do - Philipp Pointner, Jumio

Analysts point out, however, that an identification number alone won't solve every problem posed by autonomous agents. Experts note that traditional authentication methods can't keep pace with systems that operate at machine speed and make decisions without human involvement, meaning identification alone must be paired with oversight mechanisms and the ability to revoke permissions quickly.

What it means for Poland and the EU

For Poland and other European Union countries, Estonia's experiment matters for two reasons. First, it points to the direction AI agent regulation could take within the EU's legal framework, including the eIDAS 2.0 regulation on European digital identity, which already envisions digital identity wallets for citizens. Second, Poland is developing its own digital identity infrastructure around the mObywatel app, so the Estonian model could become a reference point when designing similar solutions at home.

Companies already using AI agents for customer service, accounting, or dealing with government agencies could gain legal clarity if a similar system emerges in other member states. Without such a mechanism, liability for an agent's mistakes remains legally murky, which makes it harder for companies to scale automation of processes that require access to sensitive data or financial operations.

Estonia has not yet released a rollout timeline or details on who bears responsibility when an agent makes a mistake, for example filling out a tax return incorrectly or exceeding an approved payment limit. Those questions remain open, with the government saying only that details will follow in the coming months alongside legislative work.