Estonia - National AI Strategy (2019–2021)

Overview

Estonia’s National Artificial Intelligence Strategy 2019–2021, commonly referred to as the Kratt action plan or "Kratikava", was prepared on the basis of an expert-group report and adopted by the Cabinet of the Republic of Estonia on 25 July 2019. The document functioned as a short-term, three-year national implementation plan that set out concrete measures, responsible agencies, deadlines and indicative budgets to accelerate AI uptake across the public and private sectors, strengthen skills and research and adapt the legal and governance environment to support trustworthy AI. The government committed to an initial investment envelope of at least EUR 10 million for implementation during 2019–2021 and placed the Government CIO Office (within the Ministry of Economic Affairs and Communications, MKM) in charge of steering delivery. The plan was presented publicly prior to adoption and was intended to operate as Estonia’s national AI action plan under the EU coordinated plan on AI. Key source materials include the published government action plan (full text PDF: Eesti riiklik tehisintellekti alane tegevuskava 2019–2021 (full text PDF)) and supporting documents prepared by the expert group. Many descriptions and measure-level details are summarised from the plan and supporting analysis available in the official text and related presentation materials (see links within the detailed sections below).

Definitions

The strategy adopts the local metaphor "kratt" to denote AI systems, components and reusable building blocks intended for public-sector reuse. More generally, the plan refers to AI as systems and methods that perform tasks and provide decision‑support or decision‑making functions using data, machine learning and related computational techniques. The Strategy relies on the definitions, taxonomies and scope established by the expert-group report that underpinned the action plan and uses "kratijupid" to denote reusable open-source AI components. Where the plan itself cites supporting analysis and technical definitions, those references are contained in the action-plan documentation and related annexes as published by the Ministry of Economic Affairs and Communications (MKM full text PDF) and accompanying presentation materials referenced at the time of public presentation.

Governance and Institutional Framework

Steering and coordination responsibilities for implementation were assigned principally to the Government CIO Office within the Ministry of Economic Affairs and Communications (MKM). The plan foresaw a multi-agency steering group and regular reporting to the e-Estonia Council to monitor and guide delivery. Individual measures in the plan identify lead and participating agencies and target deadlines; financing decisions and agency-level implementation responsibilities were the primary mechanisms used to drive compliance. The State Chancellery and other central government bodies were involved in coordination, while public-sector pilots and implementation activities were to be delivered by named ministries, agencies and state-funded research and training providers. The plan did not create new autonomous enforcement bodies; rather, oversight and scaling were to be conducted through existing government governance structures (steering group, e-Estonia Council and the Government CIO Office) and via integration into existing public procurement and project governance practices.

Key Focus Areas

• Advancing AI use in the public sector — promote public-sector pilots, reuse of open-source components ("kratijupid"), sandboxes for testing public-sector AI solutions and the Bürokratt initiative to modernise administrative processes.
• Supporting AI uptake in the economy — fund measures and networks to catalyse private-sector adoption, public–private cooperation and targetable incentives to increase business uptake of AI.
• Building human capital and R&D capacity — expand training and education programmes (including the Estonian translation and rollout of the Elements of AI course), create targeted training for public servants, and strengthen research and development capacity in AI across universities and research organisations.
• Adapting the legal and regulatory environment — develop guidance on responsible data use, impact assessment methodologies, procurement guidance and proposals for further legal analysis where sectoral legal gaps were identified (including alignment with data protection and administrative law principles such as GDPR).

Implementation Framework

The Strategy operates as an action plan rather than binding legislation. It sets out named measures with indicative budgets, responsible agencies and deadlines across the four pillars noted above. Implementation readiness and compliance rely on agency-level responsibilities, allocation of the committed investment funds (minimum EUR 10 million for 2019–2021), project-level governance, inclusion of AI considerations in procurement and the creation of practical instruments (guidance documents, sandbox rules, evaluation criteria) to support compliant deployments. Practical tasks for implementers included appointing data stewards within agencies, running pilots in designated sandboxes, adopting procurement guidance and establishing reporting lines to the steering group and e-Estonia Council. The plan also emphasised production and reuse of open-source AI components to reduce duplication of effort across government and encourage interoperable solutions (the "kratijupid" concept).

Monitoring and Evaluation

The action plan specified annual overviews of implementation and placed monitoring responsibilities with the steering group and Government CIO Office. Regular reporting to the e-Estonia Council and periodic aggregation of progress across named measures were intended to provide transparency on delivery. Implementation monitoring included tracking of pilots, training rollouts, roadmaps for key areas and outputs such as open-source components and the Kratt portal. The plan was explicitly framed as a foundation for successor action plans (notably the 2022–2023 cycle and later plans) that would build on outputs and lessons learned from the 2019–2021 period.

Penalties, Liability, and Appeals

The Strategy itself did not create new enforcement regimes, penalties or administrative appeals procedures. Legal and regulatory risks arising from AI deployment — for example, privacy and data-protection risks under GDPR, administrative-law constraints on automated decision-making, and liability questions — were to be addressed through follow-up legal work, development of guidance, and where necessary proposals for amendments to sectoral laws. Any enforcement or sanctions therefore remain within Estonia’s existing legislative and supervisory frameworks and applicable sectoral regulator remits.

Relationship to Other Instruments

The plan was designed to be consistent with EU-level coordinated AI activities and to operate within Estonia’s domestic legal framework, including data protection obligations under the General Data Protection Regulation (GDPR) and administrative procedure law. Where the plan identified legal gaps or areas requiring further study, it proposed follow-up analysis and potential legislative amendments. Implementation guidance, procurement rules and sandbox processes were to be developed to ensure that deployments conform with existing obligations and to reduce legal and regulatory risk for public‑sector adopters.

International Alignment

The strategy was framed as Estonia’s national contribution to EU-coordinated AI planning for the 2019–2021 period and intended to align with international best practices in trustworthy AI, including attention to privacy, accountability and human oversight. It also sought to leverage international learning in training and open-source approaches, while positioning Estonia’s outputs (open components, training platforms and pilots) to be interoperable with broader European Union efforts. The plan’s role as a national action plan under the EU coordinated plan on AI was explicitly stated in the document and in related government communications.

Implementation Timeline

Sources and References