Estonia will begin developing a digital identity for AI agents — AI-isikukood. The Eesti.ai council under Prime Minister Kristen Michal has backed the project, which is meant to allow such systems to act on behalf of a person, company or institution within limited and verifiable bounds, according to the government’s website. According to Michal, in the future artificial intelligence will more often carry out digital operations on a user’s behalf: compile reports, prepare declarations or interact with information systems. For this, it must be clear who is acting, on whose behalf, with what rights and who bears responsibility. “If we act quickly and wisely, Estonia will become the first country in the world to create an official digital identity for AI agents,” the prime minister said. Why agents need a digital ID AI-isikukood is meant to solve the problem of excessive access. At present, a user, company or institution may be forced to grant an AI assistant overly broad rights. The Estonian authorities want to replace this with limited authority. An agent will be able to act only within the granted rights: for example, view data, prepare a document, generate a payment or operate within a set limit. Michal described the concept as the issuance of “limited, controlled and auditable authority.” The launch timeline, technical architecture and the mechanism of liability for potential losses have not yet been disclosed. What Eesti.ai does The government launched the Eesti.ai initiative on January 27. Its nearest priorities were named as education, healthcare and security. The aim of the program is to systematically introduce artificial intelligence into the economy and public administration. On April 9, the Eesti.ai council approved 15 projects with an expected high impact. They concern skills, healthcare, entrepreneurship, the public sector, education and infrastructure. By June, some of them had moved into the stage of analysis, market consultations, tenders and partnership preparation. According to the authorities, the area of AI-skills training is advancing the fastest. Infrastructure The AI-isikukood project draws on Estonia’s already existing digital infrastructure. The Bürokratt website states that from 2026 each agency or subject area will be able to use its own personalized AI agent as part of a single cooperative network. Another related project is Aruait. The state information system RIA describes it as a sovereign governance layer for AI agents in the public sector. It is meant to help define the technical architecture, pilots and cooperation models for systems that act on behalf of people, companies or government bodies. According to Decrypt, the AI-isikukood idea continues Estonia’s broader line on digital identification. The country already has an electronic ID for citizens and e-Residency for foreigners. https://forklog.com/cryptorium/chto-takoe-ii-agenty-i-kak-oni-uproshhayut-zhizn-web3-polzovatelej As a reminder, in June the Anthropic team published the guide Zero Trust for AI agents on the Claude blog, covering the secure deployment of autonomous AI agents in a corporate environment. The document outlines the key risks of agentic systems and an approach to business cybersecurity.

Source: ForkLog