Estonia’s prime minister’s office announced on June 18, 2026 that the country intends to become the first in the world to issue official digital identities to artificial intelligence agents, not just to people. The proposed “AI ID codes” would let an AI system act on behalf of a person, a company, or a government agency within defined limits, while leaving a verifiable record of who authorized it and what it was allowed to do, according to reporting by Biometric Update.

The stated case is accountability, not control

The government’s own framing is narrow and procedural. “In the future, AI will increasingly carry out digital tasks on our behalf,” the prime minister’s office said. “To that end, it must be clear who is acting on whose behalf with what rights, and who is ultimately responsible.” Industry commentary quoted alongside the announcement, including from identity-verification firm Jumio, described the approach as a necessary foundation for “auditable permissions” and “user control” as AI agents take on more real-world tasks like filing paperwork or making purchases on a person’s behalf.

What the coverage so far leaves out

What is notably absent from the reporting on this announcement is any civil-liberties organization weighing in. That is not the same as there being nothing to weigh in on. A system built to track which AI agent acted on whose behalf, with what permissions, is also, by construction, a system that logs a detailed record of automated activity tied to a real identity, sitting inside a state-run identity infrastructure. Estonia already runs some of the most extensive digital-government plumbing in Europe, and every European Union member state is separately required to offer a digital identity wallet to citizens by the end of this year under the eIDAS2 Regulation. An AI-agent identity layer would not be a standalone experiment; it would plug into infrastructure that is being built out everywhere in the EU regardless of what Estonia decides.

Why this is worth tracking now, not later

The pattern worth watching is a familiar one: a government builds an accountability tool aimed at a specific, reasonable-sounding problem, before the public conversation about its broader reach has really started. That does not make the tool wrong. It does mean the actual privacy and oversight questions, who can query these records, how long they are kept, whether the same rails get extended from AI agents to tracking people directly, are being decided now, quietly, while the coverage is still limited to industry voices explaining why the idea is good.

Sources: Biometric Update, “Estonia’s AI agent ID plan raises new questions about digital identity”.