Estonia Becomes First Country to Give AI Agents Their Own ID Numbers

Estonia has announced a plan to assign artificial intelligence agents their own identification numbers, similar to those long issued to citizens under the country's e-ID system. Prime Minister Kristen Michal presented the plan as a way to solve a problem already affecting millions of AI assistant users: today, an agent booking a flight, filing taxes, or editing a document has to borrow its owner's entire digital identity, because it has none of its own.

How AI ID Codes Work

Instead of handing an agent full access to all of its owner's services and data, the system would let it be assigned narrowly defined permissions tied to a verifiable identity. An AI ID code would precisely spell out what a given agent is allowed to do: whether it may only view a record, draft a document, or authorize a payment within a set spending limit.

Every action an agent takes would leave a trace that could be checked after the fact. That sets the Estonian proposal apart from current practice in most AI systems, where an agent operates with the same rights as the user who launched it, with no separation of roles and no easy way to limit its scope of action.

It cannot be that a person is forced to give their AI assistant access to all of their rights, services, and data. Agents must have limited, controllable, and auditable permissions - Kristen Michal, Prime Minister of Estonia

Built on the e-ID Foundation

Estonia has long been seen as a leader in digitizing public administration. The country runs the e-ID system, which lets citizens handle nearly all government business online, alongside the m-Residency program, which allows foreigners to run digital companies from anywhere in the world using a mobile identity. AI ID codes are meant to be a natural extension of that same infrastructure to a new type of entity operating online.

Prime Minister Michal suggested that if the project is carried out carefully, it could become a model for other countries wrestling with the same problem: how to oversee AI agents that increasingly act on their own on behalf of people and companies, without a human checking every single step.

Open Questions

The government has not yet disclosed when the new system will take effect or which agents it will cover. It also remains unclear how Estonia intends to enforce registration for agents operating outside its jurisdiction, such as those launched by global cloud platforms based abroad. The missing details mean this is, for now, a statement of direction rather than a finished regulation.

What It Means for Other Countries

The idea arrives as AI agents capable of independently booking services, making payments, or editing documents stop being a novelty and become part of the daily work of companies and governments. The European Union is only just building its oversight framework for AI models under the AI Act, which focuses mainly on model providers rather than the identity of individual agents acting on their behalf.

The Estonian model, if implemented and proven in practice, could become a reference point for other EU member states, including Poland, which are only beginning to build their own oversight mechanisms for AI systems. The question of who bears responsibility for an autonomous agent's actions keeps resurfacing around high-profile security incidents, and the coming months will show whether other governments follow Tallinn's lead.