The Ministry of Industry and Trade has prepared a draft law on artificial intelligence, which implements Regulation (EU) 2024/1689 of the European Parliament and of the Council (hereinafter referred to as the “AI Act”). It is a minimalist draft law that does not place an unnecessary administrative burden on companies. It only regulates the necessary institutional, procedural, and sanctioning mechanisms that the AI Act entrusts to member states for independent decision-making. The draft is expected to enter into force in 2026, but no later than August 2, 2026.
The draft provides for the establishment of a regulatory sandbox for artificial intelligence, to be operated by the Czech Standardization Agency. This is a controlled testing environment that allows providers or prospective providers of AI systems to develop, test, and validate innovative AI systems for a limited time before their placement on the market or putting into service. These sandboxes may also involve testing in real-world conditions, but they will always be under the direct supervision of the relevant authority. Participation in it is based on a contract and its purpose is to increase legal certainty and support innovation and competitiveness, especially for small and medium-sized enterprises and start-ups. The requirement to operate at least one such sandbox at the national level within the member states is one of the key provisions of the AI Act, which will be applicable from August 2, 2026.
In the case of high-risk AI systems under the AI Act, these are systems that may pose a risk of harm to the health, safety, or fundamental rights of natural persons. Therefore, the strictest rules apply to them. Typically, this involves healthcare, employment, education, justice administration, critical infrastructure, etc. The draft law regulates the procedural aspects for two situations. The first is the testing of high-risk AI in real-world conditions, where the entity must first apply for approval of this testing. Regarding this application, the draft sets a specific deadline for the supervisory authorities to issue a decision in particularly complex cases of 90 days from the initiation of the proceedings. The second is the operation of high-risk AI, with a decision deadline for the application of up to 120 days.
The proposal divides the supervision of compliance with the rules among several institutions. The main institution that will supervise compliance with the obligations under the AI Act and this draft law will be the Czech Telecommunications Office. It will also act as a single point of contact for the public. Other supervisory authorities will include the Czech National Bank, which will supervise persons subject to its supervision, the Office for Personal Data Protection, the Public Defender of Rights, and the Czech Office for Standards, Metrology and Testing.
The provisions of the AI Act are based on several fundamental principles, including a risk-based approach, human oversight of key decisions, transparency, protection of fundamental rights, and ensuring the quality and safety of AI systems. Violations of the obligations established by the AI Act may result in heavy penalties, with fines of up to EUR 35,000,000 or 7% of the entity’s global turnover.
Download Legal Update 03/2026 here.