Estonia Eyes IDs for AI Agents

Every time software acts for you online, it normally wears your whole identity. It uses your login, your permissions and, in legal terms, your name. For anyone in Malta used to eID, online tax forms and government portals, the weakness is easy to grasp: it is like sending someone to Castille with your ID card, your house keys and your bank card, when all they needed was permission to submit one form.

Estonia thinks there is a cleaner way. Eesti.ai, an advisory council set up at the initiative of Prime Minister Kristen Michal, has agreed that the country should develop a separate digital identity for AI agents. Estonian media have called it an "AI-isikukood", borrowing the term for the personal identification code on which Estonia's digital state is built (ERR). The label sounds bigger than the proposal. This is not about making software a legal person. It is about stopping software from having to pretend to be you.

The digital deputy

The model has three parts. A badge says which software agent has turned up. A permission slip says what it may do, for whom, and within which limits. A receipt records what actually happened. Most AI tools today have, at best, only the first part. If an agent files your tax return or checks a public register, it often does so using your full credentials, with access well beyond the task in front of it.

Technology entrepreneur Kaspar Korjus gave a practical example: an AI agent with its own identity could prepare tax declarations for a citizen, while sensitive actions such as moving money would remain under human control (Äripäev). The useful version is a tightly limited deputy. It can read this record, draft that form, and nothing more. The citizen remains the legal principal and can revoke the delegation at any point (Global Relay).

Estonia can even discuss this because its digital plumbing already exists. Its eID system handles national identification (e-Estonia), while X-Road allows secure data exchange between services and registries (e-Estonia). An AI-agent identity would sit on top of that system. It would not replace it. For now, though, there is no law and no published technical specification. This is a direction of travel, not a working product.

The permission slip Europe hasn't written

The strongest objection is that identity is only the visible layer. Estonian commentator Peeter P. Mõtsküla put the issue plainly: the central question is not "who is the agent?" but "what may it do, and who is responsible for it?" (ERR). Claudia Plattner, Germany's federal cybersecurity chief, has warned that standards for managing AI-agent identities are still missing, especially when agents act for companies in email or online services (FAZ). German security analysts already classify AI agents as "non-human identities": machine accounts whose credentials are often poorly monitored and quietly gain wide access (Security-Insider).

The EU already has laws for proving who you are (eIDAS 2.0), limiting the data you disclose, logging AI systems (AI Act), protecting personal data (GDPR) and securing network access (NIS2). What it does not yet have is one rulebook for a software deputy that cuts across all of them. No single regulation covers the whole chain: identifying the agent, defining its mandate, recording its actions and deciding who carries the loss when things go wrong.

That last gap matters most. A digital badge does not decide who pays when delegation fails: the company that built the model, the business that deployed it, or the organisation that authorised the agent to act (FAZ). The EU's AI Liability Directive was meant to deal with this problem. It is still only a proposal (AI Liability Directive).

Whose badge, whose rules

Member states are already moving, but not in the same direction. Denmark's AI sandbox builds delegation on top of MitID, its national login system (Datatilsynet). The Netherlands publishes a public register of government AI systems, treating accountability as an administrative duty (Algoritmeregister). Sweden is working through how to certify the digital wallets foreseen under eIDAS 2.0: secure phone apps that can prove one fact about you, such as age or residency, without handing over the whole identity file (DIGG).

That fragmentation is where the real test begins. An AI agent authorised in Estonia may need to be trusted by a Danish login system, audited under Dutch accountability rules, and secured against the machine-account risks worrying German regulators. Estonia can build the badge. Europe still has to decide how that badge is checked, limited and enforced across borders.