Estonia will assign ID numbers to AI agents.

      When an AI agent acts on your behalf today, it typically impersonates you. It logs in using your credentials, which gives it access to everything. Estonia aims to change this.

      The country plans to issue personal identification numbers specifically for AI assistants, making it the first nation to do so, according to Prime Minister Kristen Michal. The concept is to provide an AI with its own identity, allowing its actions to be limited, monitored, and tracked.

      “It cannot be that people are compelled to grant their AI assistants access to all their rights, services, and data,” Michal mentioned on X. “Agents must have restricted, manageable, and auditable permissions.”

      Why a country built on digital IDs requires a new one

      Estonia is the ideal location for this initiative. Its 1.3 million citizens already utilize digital IDs for activities such as marriage, medical visits, and signing documents, and its e-Residency program provides the same digital identity to foreign entrepreneurs.

      However, that system was based on the premise that only humans possess identity and accountability. An AI agent lacks the legal capacity to authenticate, sign, or assume responsibility, meaning it can presently only borrow a user’s credentials in full.

      Assigning unique IDs to agents is intended to bridge that gap, similar to how e-Residency established a legal identity for individuals who had never been in the country.

      The agents are already integrated into government operations

      This is not a theoretical scenario for Estonia. The country has deployed AI chatbots in all schools through collaborations with OpenAI and other organizations, and operates Bürokratt, an expanding network of AI agents managing public services.

      Michal, who has an AI advisory council comprising tech entrepreneurs, recently developed a "PM Cockpit" to track governmental priorities leveraging Anthropic's Claude during a coding session. Other regions like Ukraine and Singapore are also advancing in this area, from projects like Diia.AI to licensing initiatives.

      The unresolved aspects

      Michal did not specify a start date or clarify how liability would be determined if an agent with its own ID errs. These difficult questions remain unanswered.

      Granting machines a formal identity within essential state systems is a significant accountability and security risk. Nevertheless, it shifts the narrative around agents: the objective is not to liberate them but to maintain them on a visible leash, as Europe strives to establish its own regulations regarding AI.