AI Act in Estonia: short answer
The AI Act (Regulation (EU) 2024/1689) applies directly in Estonia. It entered into force on 1 August 2024 and applies in stages. Prohibited practices and AI literacy have applied since 2 February 2025; general-purpose AI obligations since 2 August 2025; high-risk Annex III systems from 2 August 2026; and Annex I product rules from 2 August 2027.
The first step is not a policy document. It is an inventory of AI systems, their role, intended purpose and risk class. Classification drives the duties.
AI Act timeline
| Date | What applies |
|---|---|
| 1 August 2024 | The regulation enters into force. |
| 2 February 2025 | Prohibited practices and AI literacy duties. |
| 2 August 2025 | General-purpose AI model obligations, governance and penalties. |
| 2 August 2026 | High-risk obligations for Annex III systems and certain transparency duties. |
| 2 August 2027 | High-risk rules for AI embedded in Annex I products. |
Risk classes
- Prohibited: social scoring, certain manipulative techniques, untargeted facial-image scraping and other listed practices.
- High-risk: systems in Annex III (for example employment, education, essential services, border control, justice) or safety components under Annex I.
- Limited risk: transparency duties, such as informing users that they interact with AI or that content is synthetic.
- Minimal risk: most other systems; voluntary codes may still apply.
- GPAI: general-purpose models have separate transparency and systemic-risk duties.
Provider, deployer, importer or distributor?
The same technology can create different duties depending on your role. A provider develops or places an AI system on the market under its own name. A deployer uses it in a professional context. Importers and distributors have verification duties. A deployer can become a provider if it substantially modifies the system or puts it on the market under its own brand.
Write the role down before you buy or build. Procurement contracts should allocate documentation, incident notification and access to logs.
High-risk AI duties
- Risk-management system across the lifecycle.
- Data governance, representativeness and bias checks.
- Technical documentation and instructions for use.
- Record-keeping and logging.
- Transparency and human oversight.
- Accuracy, robustness and cybersecurity.
- Quality-management system and post-market monitoring.
- Serious-incident reporting and corrective action.
A penetration test can support the cybersecurity evidence. It does not replace the quality-management system, data governance or human oversight.
General-purpose AI
Providers of GPAI models need technical documentation, information for downstream providers, a copyright policy and a summary of training content. Models with systemic risk also need model evaluation, adversarial testing, incident reporting and cybersecurity measures.
If you fine-tune or integrate a GPAI model into a high-risk system, you may take on provider duties for the modified system.
What evidence to keep
- AI system inventory, owner, purpose and role.
- Risk classification and the reasoning behind it.
- Data sources, quality checks and bias testing.
- Technical documentation, model and version records.
- Human-oversight design and operating evidence.
- Logging, incident and post-market monitoring records.
- Supplier contracts, model cards and change notifications.
Request a scoping call or read the NIS2 guide if the same system is also in scope of KüTS.
Frequently asked questions
When does the AI Act apply?
The AI Act entered into force on 1 August 2024. Prohibitions and AI literacy apply from 2 February 2025; general-purpose AI obligations from 2 August 2025; high-risk Annex III obligations from 2 August 2026; and Annex I product rules from 2 August 2027.
Who supervises the AI Act in Estonia?
TTJA is the market-surveillance authority for AI in Estonia. Other regulators may have sector-specific roles.
What is a high-risk AI system?
An AI system listed in Annex I or Annex III, or one that is a safety component of a product covered by EU harmonisation law. High-risk systems carry requirements for risk management, data governance, documentation, logging, human oversight, accuracy, robustness and cybersecurity.
Do general-purpose AI models have duties?
Yes. Providers of general-purpose AI models have transparency, documentation and copyright-policy duties; models with systemic risk have additional evaluation, incident reporting and cybersecurity obligations.
Does a penetration test prove AI Act compliance?
No. Testing can provide evidence for robustness and cybersecurity, but it does not replace the risk-management system, data governance, technical documentation, human oversight or post-market monitoring.