Artificial intelligence is no longer just a technology topic — it's now a legal one too. From 2 August 2026, the main wave of obligations under the EU's Artificial Intelligence Act (AI Act) takes full effect, and it applies to ordinary companies and websites, not just major tech firms. Slovakia is meanwhile preparing a national law that will designate supervisory authorities and a domestic penalty framework.

What the AI Act is and when it applies

The AI Act is Regulation (EU) 2024/1689, the world's first comprehensive regulation of artificial intelligence. It entered into force on 1 August 2024 and, being a regulation, applies directly in Slovakia without the need for a separate national implementing act. It classifies AI systems by risk level — unacceptable, high, limited and minimal — each carrying different obligations.

What changes from 2 August 2026

From this date, Article 50 of the AI Act on transparency applies. Companies and website operators must in particular ensure that:

  • chatbots and other systems interacting with people are designed so it is clear the user is dealing with AI, not a human,
  • content generated by AI (text, image, audio, video) is marked in a machine-readable format as artificially created,
  • deepfakes and AI-generated text on matters of public interest are visibly labelled, unless they have undergone genuine human editorial review,
  • the use of emotion recognition or biometric categorisation systems is disclosed to the people affected.

A separate obligation has applied since February 2025 — companies must ensure a sufficient level of AI literacy among employees who work with artificial intelligence. Internal training, a usage policy, and documentation of these measures are sufficient.

High-risk systems and machine-readable marking are being postponed

Under the so-called Digital Omnibus, some of the originally planned deadlines have been pushed back. Obligations for standalone high-risk systems (e.g. AI used in recruitment or credit scoring) will not apply from August 2026, but only from 2 December 2027. Machine-readable marking (watermarking) of AI content for systems placed on the market before August 2026 is postponed to roughly December 2026. The remaining transparency obligations (chatbot disclosure, deepfake labelling), however, still apply from 2 August 2026.

What penalties are at stake

Sanctions under the AI Act are among the strictest in European legislation:

  • up to €35 million or 7% of worldwide turnover for prohibited practices (e.g. social scoring of citizens),
  • up to €15 million or 3% of worldwide turnover for breaching other obligations, including transparency,
  • up to €7.5 million or 1% of worldwide turnover for providing incorrect or misleading information to the authorities.

The higher of the two amounts always applies, while for small and medium-sized enterprises the lower amount applies instead. Fines are imposed by the national supervisory authorities of the individual member states.

Slovakia's AI law — still only a draft

Slovakia is preparing an act on artificial intelligence and European data regulation, which is meant to designate supervisory authorities, their powers, and a Slovak penalty framework. The Ministry of Investments, Regional Development and Informatization (MIRRI) submitted an integrated draft in early 2026; the comment procedure generated hundreds of comments, and the draft is now under review. The key open question is which body will act as the main supervisory authority — either MIRRI itself as a single point of contact, or a newly established Office for Digital Integrity, alongside the Office for Personal Data Protection, the National Security Authority, and various sectoral bodies. Until the act takes effect, only the EU regulation applies directly.

What companies should do now

  • Draw up a list of the AI tools your company actually uses (chatbots, content generators, recommendation systems, filters, translation tools) and classify them by risk level.
  • If your company runs an AI chatbot, make sure there is a clear and visible notice, right from the first interaction, that the user is dealing with AI.
  • For published AI-generated content (marketing, images, video), consider a visible label, especially for realistic-looking outputs.
  • Introduce an internal AI usage policy, including a ban on entering sensitive client data into publicly available tools.
  • Train employees who work with AI and document that training (the AI literacy obligation).
  • Keep track of the legislative progress of the Slovak act and prepare for it as it develops.

If you need help assessing which risk category the AI tools used by your company fall into, or preparing an internal policy and documentation under the AI Act, we are happy to help.