Publication date: December 09, 2024
The government’s draft law on artificial intelligence systems “serves to apply Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 on the establishment of harmonised rules on artificial intelligence [1]“, also known as the AI Act . This document defines the principles of supervision of the market of artificial intelligence systems and the AI model and covers the organisation of supervision, the conditions for accreditation of conformity assessment bodies, incident reporting and the imposition of administrative penalties for the use of prohibited practices under Article 5 of the European Regulation on artificial intelligence. Moreover, this draft is to be applied, among others, to persons located in the EU affected by AI, importers and distributors of AI systems, entities using AI systems that are established or located in the Union, and suppliers introducing AI systems to the market or putting them into service [2]. Additionally, the act provides for the creation of the Office of the Commission for the Development and Security of Artificial Intelligence, which is a state legal person [3].
The AI Development and Security Commission acts as the central authority for the supervision, development and regulation of AI in Poland. According to Article 5 of the draft act, the AI Development and Security Commission is served by the aforementioned Office of the AI Development and Security Commission, which in turn is supervised by the Prime Minister[4].
“The Commission is the market surveillance authority for artificial intelligence systems” [5]. Its activities will focus on promoting innovation and supporting AI research and applications in various sectors, including by organising regulatory sandboxes [6] that will enable testing of technologies in controlled conditions. According to Article 11 of the discussed draft, the Commission’s tasks include monitoring compliance with the provisions of Regulation 2024/1689 (AI Act ) and the Act, taking actions to ensure the proper functioning of the internal market of the European Union. An important aspect of the Commission’s activities is also counteracting threats related to the security of artificial intelligence systems, including receiving reports of serious incidents related to the use of these systems. Moreover, this body also participates in the development and opinion on draft legal acts on artificial intelligence and issues resolutions and decisions in cases of infringement of the provisions on artificial intelligence systems. Its tasks also include exercising the competences of the market surveillance authority, specified in Regulation 2024/1689 (AI Act ), as well as developing and issuing publications and implementing educational programs aimed at popularizing knowledge about artificial intelligence. The Commission also conducts information activities and implements Poland’s international obligations in the field of cooperation and exchange of information regarding the supervision of artificial intelligence systems.[7] Additionally, “the Commission may develop explanations, opinions and interpretations that are of significant importance for the application of regulations in matters covered by the scope of the Commission’s activities. Explanations, opinions and interpretations are published in the Public Information Bulletin on the website of the Commission’s Office[8].”
According to the draft act on artificial intelligence systems, the Commission is a collegial body and its composition “consists of the Chairman of the Commission, two Deputy Chairmans of the Commission and nine members of the commission[9]“, who are representatives of:
1) the minister responsible for computerization;
2) the minister appointed to coordinate the activities of the special services and, if no minister has been appointed, the Chancellery of the Prime Minister;
3) President of the Personal Data Protection Office;
4) President of the Office of Competition and Consumer Protection;
5) the Polish Financial Supervision Authority;
6) the Commissioner for Human Rights;
7) the Children’s Rights Ombudsman;
8) the National Broadcasting Council;
9) President [10] of the Office of Electronic Communications.
The Chairman’s task is to manage the work of the committee and represent it externally [11], and the members of the committee should support him in performing the statutory tasks [12]. Moreover, the Chairman’s duty is to appoint a secretary, who notifies the members of the Committee of the dates of its meetings, draws up minutes of them and stores their documentation. [13]
The process of selecting members of the Commission and their terms of office are specified in detail in the draft law, highlighting the requirements regarding their qualifications, independence and work ethic.
The Chairman of the Commission is elected for a five-year term by the Prime Minister through an open and competitive recruitment process. Candidates must meet strict requirements, including Polish citizenship, full public rights and at least a master’s degree or equivalent. They should also have significant achievements in the field of artificial intelligence, law or related technological fields, and at least three years of experience in a managerial position. Additionally, candidates must enjoy an unblemished reputation, not be convicted of intentional crimes and provide a guarantee of proper performance of the assigned tasks. One person may hold the position of Chairman for a maximum of two terms.[14]
The Deputy Chairmen of the Commission are appointed by the Prime Minister at the request of the Chairmen of the Commission from among persons indicated by him. Candidates for these positions must meet the same requirements as for the Chairmen. The Deputies support the Chairmen in the implementation of the Commission’s tasks and may perform his duties in the event of his absence.[15] The participation of representatives of these institutions provides the Commission with a broad perspective and access to expert knowledge in key areas related to AI, such as personal data protection, consumer rights, financial regulations or telecommunications supervision.
The Commission also has a Social Council for Artificial Intelligence, which serves as an advisory and consultative body. The Council consists of 9–15 members elected for a two-year term. Council members must have specialist knowledge in the field of artificial intelligence, computer science, new technology law, personal data protection and human rights. Candidates for Council members may be nominated by, among others, the Commissioner for Human Rights, the Commissioner for Children’s Rights, chambers of commerce, foundations and associations working for the development of AI, as well as trade unions.[16]
Chapter 3 of the draft act on artificial intelligence systems concerning the control of compliance with the provisions on artificial intelligence specifies the supervision mechanisms and procedures related to the verification of compliance of entities’ activities with the provisions. The Artificial Intelligence Development and Security Commission has the right to control the activities of entities obliged to comply with the provisions of the Act and Regulation 2024/1689 (AI Act ). The purpose of the control is to determine the compliance of the entities’ activities with the provisions.[17] Moreover, it can be carried out remotely using electronic means of communication, and in justified cases on-site at the headquarters of the controlled entity. It is also[18] worth noting that the controllers have the right to access documents, IT systems, rooms and to request explanations and take notes from materials related to the subject of the control. Additionally, the Artificial Intelligence Development and Security Commission has the right to secure documents and evidence, including electronic devices and data carriers. Objects can be stored in separate rooms of the entity or in the Office of the Commission[19].
As can be seen, the draft act on artificial intelligence systems is a comprehensive attempt to implement the regulations resulting from the EU Regulation 2024/1689 (AI Act ), while ensuring the adaptation of the national legal order to the requirements of modern technology. The introduction of the Commission for the Development and Security of Artificial Intelligence as a central supervisory body emphasizes the importance of a strategic approach to the management and regulation of artificial intelligence in Poland.
With specific rules for control, compliance verification and accreditation, the draft law guarantees the effectiveness of enforcement and the safety of AI system users. Creating space for innovation development, such as regulatory sandboxes and the involvement of the Social Council for Artificial Intelligence indicates balancing legal rigours with the promotion of innovation.
The introduction of this act can contribute to increasing trust in AI technology, while supporting the development of the digital economy in Poland. This is another step towards strengthening Poland’s position as a country supporting the safe, responsible and sustainable use of artificial intelligence.
[1]Text taken from the government draft act on artificial intelligence systems of October 15, 2024.
[2]The scope of application of the Act is taken from REGULATION (EU) 2024/1689 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 13 June 2024, also known as the AI Act .
[3]Art. 5 of the draft act on artificial intelligence systems of 15 October 2024.
[4]Art. 5 of the draft act on artificial intelligence systems of 15 October 2024.
[5]Art. 6. section 1 of the draft act on artificial intelligence systems of 15 October 2024.
[6]For the purposes of the Draft Act on Artificial Intelligence Systems of 15 October 2024 and REGULATION (EU) 2024/1689 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 13 June 2024, an “AI regulatory sandbox” means a controlled framework established by a competent authority that enables providers or potential providers of AI systems to develop, train, validate and test – where appropriate under real-world conditions – innovative AI systems, based on a sandbox roadmap, within a limited time period and under regulatory supervision.
[7]Art. 11 of the draft act on artificial intelligence systems of 15 October 2024.
[8]Art. 13 of the draft act on artificial intelligence systems of 15 October 2024.
[9]Art. 7 of the draft act on artificial intelligence systems of 15 October 2024.
[10]Art. 7 of the draft act on artificial intelligence systems of 15 October 2024.
[11]Art. 8 of the draft act on artificial intelligence systems of 15 October 2024.
[12]Art. 9 of the draft act on artificial intelligence systems of 15 October 2024.
[13]Art. 10 of the draft act on artificial intelligence systems of 15 October 2024.
[14]Art. 18 of the draft act on artificial intelligence systems of 15 October 2024.
[15]Art. 21 of the draft act on artificial intelligence systems of 15 October 2024.
[16]Art. 28 of the draft act on artificial intelligence systems of 15 October 2024.
[17]Art. 41 of the draft act on artificial intelligence systems of 15 October 2024.
[18]Art. 43 of the draft act on artificial intelligence systems of 15 October 2024.
[19]Art. 43 of the draft act on artificial intelligence systems of 15 October 2024.