How Estonia plans to control AI agents through digital identities

Estonia is positioning itself as the world's first country to regulate artificial intelligence through digital identities for AI agents. Prime Minister Kristen Michal approved the plan this week, creating what officials call "AI ID codes", unique digital identities that grant AI assistants limited, controllable powers instead of blanket access to all rights, services, and data.

Currently, individuals and organisations must often grant AI assistants complete access to accounts, files, and financial systems.

Estonia's digital IDs flip that model: an AI agent can be authorised to view data only, prepare specific documents, approve payments up to defined limits, or operate within restricted timeframes.

The Estonian solution takes advantage of decades of investment in digital government technology, such as the X-Road (a platform for exchanging data), digital signatures, and digital footprints. These technologies enable the trust environment that makes it possible for AI agents to use IDs safely.

"To that end, it must be clear who is acting on whose behalf with what rights and who is ultimately responsible," Michal explained. The accountability framework guarantees that whenever an AI agent does something, that action can be linked to the human behind it.

A business can use an AI agent to handle invoice processing but restrict its capability to those payments below a specified level. An individual can delegate an AI agent to do tax documents but without having access to personal financial information.