Under a new legislative proposal, the Bulgarian government would require developers to surrender source code and model parameters for any AI system deemed a threat to national security. This assertive stance reflects a growing trend among European nations to treat high-level algorithms as critical infrastructure rather than just commercial software products. By formalizing this draft law, the Bulgarian administration is attempting to align its domestic legal framework with the European Union AI Act while simultaneously asserting its unique national interests. The proposal is currently navigating parliamentary committees, where it is being analyzed for its potential to safeguard fundamental human rights in an increasingly automated society. The government is not simply adopting foreign standards; it is synthesizing them with local needs to create a stable environment for digital growth. This approach ensures that the country remains a proactive participant in the regional tech landscape throughout the 2026 legislative cycle, setting a clear trajectory for future innovation and technological safety.
Administrative Oversight: Implementing Comprehensive Regulatory Structures
The proposed framework introduces a sophisticated multi-tiered governance structure designed to ensure maximum accountability across all sectors of the economy. At the helm of this system sits the Minister of Innovation and Digital Transformation, who is tasked with serving as the primary architect of the nation’s technological policy. To bridge the gap between policy creation and day-to-day enforcement, the government intends to appoint a National Coordinator for AI. This individual would be responsible for overseeing the National AI Strategy and delivering annual risk assessments directly to the Prime Minister. Furthermore, the draft law identifies ten distinct national competent authorities that will manage market surveillance and technical compliance. The Communications Regulation Commission is slated to serve as the single point of contact for European Union coordination, ensuring that Bulgaria’s regulatory actions remain consistent with continental mandates while addressing specific local challenges within the telecommunications and data sectors.
Beyond administrative structures, the legislation establishes strict ethical boundaries to protect the social and judicial integrity of the state. One of the most significant provisions is the absolute ban on real-time remote biometric identification in public spaces, a move designed to prevent the emergence of pervasive surveillance states. Additionally, the proposal explicitly bars artificial intelligence from being utilized in the courtroom to interpret legal facts or apply abstract principles to specific cases. This preserves the essential human element in the judiciary, preventing algorithms from making life-altering legal determinations. The draft also addresses the rise of predatory economic practices by prohibiting AI-driven recommendation systems in the online gambling industry, which often target vulnerable populations. These prohibitions extend into the realm of national intelligence, where the use of AI is restricted to the preparation of evidence rather than autonomous decision-making, thereby ensuring that specialized state operations remain under human control.
National Sovereignty: Protecting Security and Economic Growth
A central theme of the new framework is the concept of digital sovereignty, which the government defines as the capacity to make independent technological choices without undue foreign influence. To achieve this, the Council of Ministers is granted the authority to classify specific AI systems as high-risk if they impact national security or critical infrastructure. Once a system is classified in this manner, operators are legally bound to provide the State Agency for National Security with comprehensive access to their technical documentation and source code. Beyond security, the state has introduced a multiplier effect for AI investments, where capital allocated to development is recognized at five times its face value for promotional purposes. Small and medium-sized enterprises are also prioritized through targeted financing and access to high-performance computing resources. These combined measures aim to reduce dependencies on non-allied technology providers while simultaneously strengthening the economic resilience of the nation’s burgeoning digital ecosystem.
The protection of public discourse was addressed through a durable ban on unauthorized deepfakes, extending fifty years beyond a person’s death while permitting a narrow window for non-defamatory satire. This focus on ethical integrity coincided with the launch of a National AI Regulatory Sandbox, which allowed developers to test innovative administrative tools under controlled conditions to ensure electronic services met safety standards before deployment. Civil servants participated in mandatory training programs to build necessary technical skills, and public data sets were systematically anonymized to provide safe training material for local models. Moving forward, organizations must prioritize the auditing of their current algorithmic assets to ensure full compliance with the impending transition window. Stakeholders should also focus on developing transparent data governance policies that exclude sensitive information regarding minors to avoid severe penalties. These proactive measures were designed to transform the public sector into a modern and secure infrastructure that serves the interests of every citizen.
