Estonia wants to become the first country in the world to grant digital identities to AI agents. The Estonian government is studying the creation of AI identification codes so that intelligent agents can act legally on behalf of people and companies under defined and auditable limits.
The government of the Baltic country has announced that it will move forward with the development of digital identities for artificial intelligence agents. This initiative could make it the first state in the world to establish an official system for AI to act on behalf of citizens, companies and organizations within a regulated and verifiable framework.
The proposal was endorsed during the second meeting of the advisory council of Eesti.ai, created on the initiative of Estonian Prime Minister Kristen Michal. According to the government, the goal is to develop a solution that allows AI agents to operate with limited and controlled powers, always maintaining traceability of the actions carried out and the ultimate responsibility of the people or entities they represent.
"In the future, AI will perform more and more digital tasks on our behalf, such as drafting reports, preparing declarations or interacting with information systems!," Michal said. "To that end, it must be clear who acts on whose behalf, with what rights and who is ultimately responsible."
A digital identity for each AI agent
The initiative provides for the creation of what the government calls "AI identification codes," a kind of official digital identity that would allow specific permissions to be assigned to intelligent agents.
In this way, a person or company could authorize an AI to carry out certain actions without granting it full access to all their data or digital services. For example, an agent could have permission only to consult information, prepare administrative documentation, make payments or execute financial operations within a previously established limit.
Why an AI token has nothing to do with a Blockchain one?
According to the Estonian executive, the proposal aims to avoid one of the main risks associated with the massive use of intelligent assistants: the need to grant excessive permissions to automated systems so that they can operate effectively.
Estonia has for years been considered one of the most advanced countries in government digitalization. Its technological infrastructure includes tools such as national digital identities, electronic signatures, the X-Road interoperability platform and advanced digital auditing systems.
For Michal, the arrival of AI agents poses a challenge similar to the one the country faced during the construction of its e-government.
The success of Estonia's digital state has been based on trust, the prime minister said. "Digital identities, the X-Road platform, digital signatures and the digital footprint have made our country faster, simpler and safer. Now that we are in the era of AI agents, we face the same question: how can we use this technology to make life easier without losing control or responsibility?", he said.
The leader assured that, if the country acts quickly, Estonia could become the first nation to implement official digital identities for artificial intelligence agents.
Drones, defense and physical AI
During the advisory council meeting, projects related to so-called physical AI and dual-use technologies were also discussed.
Among them, the "Merepesa" proposal stood out, an initiative aimed at creating a testing environment in the Gulf of Finland and the Baltic Sea for testing aerial, maritime and underwater drones.
The project aims to simplify testing in open sea, a process that is currently complex and costly due to regulatory limitations and the scarcity of specialized infrastructure. According to the government, the platform would allow simultaneous testing of autonomous systems in the air, on the water and beneath it, strengthening Estonia's regional position in sectors such as defense, robotics and unmanned vehicles.
The advisory council also reviewed the progress of several projects launched in April. Among them, the "Most AI-Expert Nation" initiative stands out, aimed at improving the population's skills in artificial intelligence.
Between April and June, a pilot phase was developed consisting of 35 practical workshops held in six cities across the country, with the participation of some 1,200 people. The goal now is to expand the program to reach 10,000 participants before the end of 2026 through a broader network of trainers and specialized courses.
Estonia also wants to protect its language in the AI era
Another topic analyzed was the classification of large language models (LLMs) developed by the Institute of the Estonian Language. The tool evaluates the performance of AI models in the Estonian language, their knowledge of national culture and their resistance to disinformation and propaganda campaigns.
According to the institute, the evolution of artificial intelligence in Estonian is not guaranteed on its own. Experts warn that cheaper models tend to offer worse results in minority languages and that the use of inadequate systems can increase the spread of erroneous information.
For this reason, the authorities consider it necessary for Estonia to participate actively in the development of models adapted to its language and culture, guaranteeing the availability of quality data for the training and improvement of future generations of AI.