Estonian government sends AI agent code draft to parliament
The proposed identifier would link an agent to a specific person without granting rights or automatic access to data or services.
Estonia’s government has approved a draft act that would create a unique “agent code” for an artificial intelligence agent acting on behalf of a named person, the Ministry of Economic Affairs and Communications said. The draft now moves to the Riigikogu. Agent codes could be introduced only after parliament adopts the act and the necessary technical solution is completed.
The code would let a service provider verify whose behalf the software is acting on. It would not give the AI agent independent rights or automatic access to the person’s data or services. That distinction matters for firms building agent-enabled services: the proposed identifier would support attribution, but would not itself grant the agent rights or automatic access to data or services.
Voluntary for companies
Companies would decide whether to support the agent code. In state digital services, an agent could use the code only where the relevant authority permits such actions.
The Information System Authority, or Riigi Infosüsteemi Amet, is developing criteria for the secure use of software solutions in state information systems and will issue implementation recommendations. The ministry says the draft’s security requirements were refined after feedback from public authorities and sector experts.
Economic Affairs and Industry Minister Liina Vahtras said, in translation, that the code should help ensure people can use AI agents “with confidence that their data is protected and that the actions taken are verifiable”. The ministry presents the measure as applying to AI-agent use in both public and private sectors, while leaving adoption by companies voluntary.
Narrower than the earlier identity plan
The government’s 24 September announcement described work on a digital identity for authorised AI agents and initial pilots in e-services. The ministry now says the draft abandons its initial digital-identity approach, and more clearly limits the agent code’s use and the Information System Authority’s duties.
In Atlas’s view, the proposal is primarily an accountability layer rather than a permissions framework. Service providers would gain a way to identify the person represented by an agent, but would still need their own rules on whether that agent may perform the requested action.
The draft’s next concrete step is its submission to the Riigikogu. Implementation cannot begin until the act is adopted and the technical solution is ready.
Permanent link to this Atlas analysis
Related Atlas analysis
Continue with coverage connected by market, topic and operating context.
- United StatesNew York AG creates ASSET Unit to enforce AI safety lawsThe unit will oversee public safety disclosures and incident reports from large frontier AI developers when the RAISE Act takes effect on 1 January 2027.
- LiechtensteinFMA plans 1 May 2027 REST switch for MiFIR reportsTesting runs from 1 February to 30 April, while ISO 20022 formats and FMA-issued certificate authentication remain unchanged.
- LiechtensteinLiechtenstein DPA warns ChatGPT ads may expose users’ intimate informationOpenAI’s planned Europe rollout would use chats, usage and device data for opted-in ads, while generic advertising could still use current chat context.
Continue with Atlas
Move from this development into the relevant research, comparison and workflow.