Estonia becomes first country to give AI agents official digital identity codes
Businesses can restrict AI agents to payments below a set limit using the new framework
Estonia is set to become the first country in the world to regulate artificial intelligence through dedicated digital identities for AI agents, after Prime Minister Kristen Michal approved the framework this week. The system introduces what officials have termed "AI ID codes" — unique digital identities that grant AI assistants specific, controllable powers rather than unrestricted access to accounts, data and financial systems. The move positions Estonia, long regarded as a global leader in digital governance, at the forefront of AI regulation.
How AI ID codes work
The new framework directly addresses a significant gap in how AI tools are currently deployed. At present, individuals and organisations typically must grant AI assistants blanket access to their accounts, files and financial systems in order for them to function.
Estonia's AI ID codes reverse that model entirely, allowing each agent to be authorised with precisely defined permissions — such as viewing data only, preparing specific documents, approving payments up to a set financial limit, or operating within a restricted timeframe.
Practical applications illustrate the system's flexibility. A business could deploy an AI agent to handle invoice processing while restricting it to transactions below a specified value. An individual could authorise an AI agent to manage their tax documentation without granting it access to broader personal financial information.
Built on decades of digital infrastructure
The framework draws on Estonia's existing investment in digital government technology, including X-Road — the country's platform for secure data exchange — as well as established digital signature and digital audit trail systems. These tools create the trust environment that makes it technically feasible for AI agents to operate with assigned identities safely and verifiably.
Prime Minister Michal was clear about the accountability principle underpinning the system. "To that end, it must be clear who is acting on whose behalf with what rights and who is ultimately responsible," he said. The framework is designed to ensure that every action taken by an AI agent can be traced back to the human responsible for it.
