Estonia becomes first nation to issue digital identities for AI agents with personal ID codes

Estonia Introduces Personal Identification Codes for AI

Estonia plans to become the first country to issue personal identification codes for AI agents, a move that addresses growing concerns about accountability and control in an increasingly automated world1

1

. Prime Minister Kristen Michal announced the initiative, which is backed by the country's Eesti.ai advisory board, stating that AI agents will receive their own digital identity separate from the humans, companies, or institutions they serve4

4

. The system will enable AI agents to perform tasks like compiling reports, preparing declarations, and interacting with information systems while maintaining clear boundaries around what they can access and do5

5

.

Solving the Access Problem with AI Agents with Digital IDs

The initiative tackles a fundamental problem in how AI agents currently operate. When an agent books a flight, files taxes, or edits a document today, it typically must borrow its owner's entire digital identity to function2

2

. "It cannot be the case that a person is forced to give their AI assistant access to all of their rights, services, and data," Michal wrote on X3

3

. The new digital identity system will provide limited, controllable and auditable authorizations, allowing users to specify whether an agent may only view data, prepare a document, or act within a fixed monetary limit1

1

.

AI Agent Accountability and the Question of Liability

The move toward AI agent governance comes as businesses grapple with agents that occasionally go rogue or misinterpret instructions, sometimes deleting entire company databases or leaking sensitive client data3

3

. An unsupervised AI agent recently ran up a $6,531 AWS bill in under a day after its owner asked it to scan a hobbyist network with no review4

4

. The question of AI agent liability remains complex. As Georgia Institute of Technology professors Mark Riedl and Deven Desai noted in their paper "AI Agents and the Law," while agency law disciplines human agents by imposing financial and even criminal penalties, "that is not so for software agents"1

1

. Courts have already begun addressing these issues: a Canadian court held Air Canada liable for bad chatbot advice, and a German court held Google liable for inaccurate AI Overview content1

1

.

Building on Estonia's Digital Infrastructure

Estonia's position as one of the world's most digitized states makes it uniquely suited to pioneer this approach to AI regulation. The country's 1.3 million residents already use digital IDs to marry, see doctors, and sign documents, while its e-Residency scheme extends the same digital identity to non-resident entrepreneurs abroad2

2

. By December 2024, Estonia had moved 100% of government services online4

4

. The country has already deployed AI agents inside government systems through Bürokratt, a state-created AI-based digital assistant that helps institutions deliver customer service4

4

. Estonia has also put AI chatbots in every school through partnerships with OpenAI and others2

2

.

The Agentic Future and International Standards

Michal framed the initiative as preparation for an agentic future that is rapidly approaching. "In the future, artificial intelligence will carry out digital actions on behalf of a person, company, or institution," he stated, adding that "it must be clear who is acting, on whose behalf, with what rights, and who is responsible"4

4

. The prime minister suggested the program could become an "international standard" for the regulation and monitoring of AI systems if executed wisely3

3

. Private sector companies are already taking positions on this issue. Target Corporation revised its Terms & Conditions with a section on Agentic Commerce, stating that "purchases and other actions taken by an Agentic Commerce Agent that you have authorized are considered transactions authorized by you." American Express took the opposite approach, announcing it would protect eligible customers from charges related to AI agent error1

1

. While Michal provided no start date for the system and no detail on how liability would work when an agent with its own ID makes a costly mistake, the initiative represents a significant step toward establishing auditable authority for autonomous systems in digital governance4

4

.