Bulgaria Prepares National AI Law as EU Enforcement Begins
By Suad Seferi ·
Bulgaria is close to completing a new national law on artificial intelligence, with the draft expected to move to public consultation before being submitted to parliament. Innovation and Digital Transformation Minister Ivan Vasilev told Bulgaria’s parliamentary Committee on Innovation and Digital Transformation on September 2 that work on the legislation was in its final stages. The government expects the text to be among several digital laws put before parliament before the end of 2026. The timing is significant. Most of the European Union’s AI Act became applicable on August 2, 2026, shifting the focus from preparing for the regulation to actually enforcing it. Although the EU AI Act is directly applicable across member states, national governments still have important responsibilities. They need competent authorities capable of supervising AI systems, applying the rules locally and coordinating enforcement with the European Commission’s AI Office. Bulgaria’s forthcoming legislation is part of that national implementation layer. Details of the Bulgarian bill have not yet been published, meaning it is too early to say exactly how Sofia intends to divide responsibilities among regulators or whether the draft will introduce additional national provisions. Those questions should become clearer once the public consultation opens. From European rules to national enforcement The EU AI Act entered into force in August 2024, but its requirements have been introduced gradually. Rules covering prohibited AI practices and AI literacy began applying in February 2025. Requirements for providers of general-purpose AI models followed later that year. The largest enforcement step came on August 2, 2026, when transparency requirements and a wider group of provisions became enforceable. National authorities and the European AI Office now share responsibility for supervising different parts of the framework. Some rules still have later deadlines. Requirements covering certain high-risk systems will come into force progressively through 2027 and 2028. This staged approach means member states are simultaneously enforcing parts of the AI Act while still building the institutions and national legal frameworks needed for the later phases. Bulgaria is now moving through that process. Under the European framework, member states are required to establish or designate national authorities responsible for market surveillance and notification. These bodies need expertise not only in AI but also in areas including cybersecurity, data protection, fundamental rights and product safety. How Bulgaria structures those responsibilities will therefore be one of the most important parts of the upcoming draft. AI is also entering Bulgaria’s public administration The legislative work comes as the Bulgarian government is separately exploring where AI can be used inside public administration. Speaking before the same parliamentary committee on September 2, Vasilev said his ministry was working with other government departments to identify duplicated processes that could be automated, including through AI. He said the current plans were not intended to replace public-sector employees with AI systems. That distinction will become increasingly relevant as governments across Europe move from discussing AI to deploying it. Public administrations can use AI for document processing, information retrieval, citizen services and other repetitive tasks. But government use also brings questions around accountability, data protection and the role of human decision-making. The regulatory framework developing alongside those deployments will determine how many of those systems are permitted and what safeguards apply. A useful test for the wider region Bulgaria’s approach is also worth watching from outside the EU. Western Balkan countries preparing for closer alignment with European digital regulation will eventually face many of the same questions: who supervises AI systems, how regulators acquire technical expertise, how complaints are handled and where responsibility sits when government agencies deploy AI themselves. For Bulgaria, those questions are moving from policy discussions into legislation. The next important step will be publication of the draft AI law. That will reveal how Sofia intends to turn the EU-level framework into a functioning national enforcement system.