A draft Act on the use and development of artificial intelligence (“AUDAI”) has been submitted to the National Assembly, with the main aim of introducing the rules of Regulation (EU) 2024/1689 (“EU AI Act”) at national level and establishing a comprehensive regulatory framework for the safe, ethical, and innovative implementation of artificial intelligence (“AI”).

The draft act is the first attempt for the adoption of a Bulgarian AI Act that not only follows the European regulation, but also adds specific national structures and mechanisms for oversight, innovation, and security. It aims to provide predictability and clarity for businesses and the public sector, while laying the foundations for a Bulgarian model for AI governance – with a focus on trust, transparency, and protection of the public interest.

The legislative initiative sets out specific rules for the public and private sectors, as well as several key elements that define the archetype of the new regulatory framework.

Firstly, it introduces the so-called National AI Coordinator, who will lead the development of a national strategy in this area, as well as a roadmap for its implementation. In addition, it provides for the creation of a network of competent supervisory authorities, including the Commission on Protection of Competition (“CPC”), the Bulgarian National Bank (“BNB”), Council for electronic media (“CEM’), Commission for personal Data Protection (“CPDP”), the Inspectorate to the Supreme Judicial Council, etc. These institutions will supervise the respective sectors, with new functions related to risk assessment and control of AI systems being added to their traditional powers.

There are also plans to establish a Consultative Council to the Council of Ministers, which will bring together representatives of business, academia, and the civil sector.

National authorities, for their part, will facilitate the entry of Bulgarian AI products into the European market and vice versa, which will provide a new level of confidence for investment and development in Bulgaria.

The second significant innovation is the creation of registers and laboratories that will serve to track and test AI solutions. There are plans to create a Register of Conformity Assessment Bodies and a Register of High-Risk AI Systems Used in Critical Infrastructure, as well as to establish a Regulatory Laboratory for testing systems in a real environment, which will support the development and implementation of safe solutions.

For businesses and start-ups, this will mean a degree of predictability – access to advice, information, and administrative support through a single help centre for support and use of artificial intelligence.

Third, the draft act encourages the use of open-source AI models for general purposes (e.g., large language models such as ChatGPT), free-to-use systems trained solely on publicly available data and without licensing restrictions. These models are seen as a tool for research and development and are expected to contribute to the development of local AI solutions, including in Bulgarian language.

The use of public data is equated with processing in the public interest, provided that anonymity and protection of personal information are guaranteed.

Among the most important safeguards in the draft act are prohibitions on high-risk applications such as remote biometric identification, the use of AI for manipulative or misleading content, deep-fake content without consent, or the application of AI in special intelligence means. Practices that violate fundamental rights and principles of non-discrimination are also prohibited.

Serious penalties are envisaged for violators, such as fines of up to BGN 5 million or up to 5% of the annual turnover of the operator of a very large online platform.

In economic terms, the draft act focuses on promoting innovation. AUDAI provides for state support measures - provision of computing resources, targeted funding, access to public data, as well as incentives under the Investment Promotion Act and funds from European funds.

For the first time, a requirement for systematic training of civil servants and public sector employees in working with AI technologies is introduced, which is seen as part of a long-term digital transformation.

However, the draft act is not limited to the framework of innovation. It also includes mechanisms to protect national security and ethical standards, including transparency requirements for systems that may pose security threats, control over the recommendation algorithms of large online platforms, and a ban on the use of children’s personal data for model training, except for educational purposes.

From an enforcement perspective, key issues include the definition of “high risk”, which models are subject to assessment, what technical and organizational measures are required, and how compliance is ensured. It will be important for companies to understand when their AI solutions fall within the scope of this law, as well as what obligations they have in terms of transparency, accountability, oversight, and personal data protection.

For businesses, the new legislative framework is both an opportunity and a challenge, as it promises predictability and clarity by facilitating access to public resources and data in Bulgaria. At the same time, it requires compliance with high standards from the outset – significant administrative and technical commitment – through registration, documentation, conformity assessments, and compliance with ethical requirements. Startups and small companies will have to strike a balance between compliance costs and opportunities for market access and trust.

The draft act has yet to be passed, but its introduction marks an important moment for technology companies, AI developers, and the entire regulatory ecosystem in Bulgaria. In the coming months, it will be crucial to find a balance between incentives and restrictions, as well as how institutions will implement the new requirements. The success of the new framework will depend on proportional enforcement and the state’s willingness to support innovation without hindering it. The draft act places Bulgaria among the first countries in the EU ready to implement the EU AI Act with national specificity, combining regulatory control with real mechanisms to stimulate business, technological development, and trust in ethical AI.