Starting August 2, 2025, the European Union enforced new regulations targeting general-purpose artificial intelligence (GPAI) models - the technology behind tools like ChatGPT, Copilot, and other generative AI applications. These rules form a key part of the AI Act, aimed at ensuring responsible development and deployment of such AI systems.

Under these regulations, developers of GPAI models must maintain and update technical documentation, provide clear and accessible information for organizations integrating these models into their products and services, comply with copyright laws, and publish summaries of the training data in a format approved by the European Commission.

Beyond these baseline obligations, providers whose models pose systemic risks, such as potential discrimination affecting large population groups, are required to conduct formal model assessments using standardized protocols. They must also identify, mitigate, and monitor risks at the EU level, report incidents to the AI Incident Office, and maintain robust cybersecurity measures to protect users and systems.

In July, the European Commission introduced a voluntary Code of Conduct for General-purpose Models, designed as a guideline to foster ethical and responsible AI practices. This Code addresses transparency, copyright, security, and risk management, offering companies and organizations a framework to prepare for future legal obligations while building trust and safeguarding fundamental rights across the AI supply chain.

Playing a significant role in shaping these emerging standards is the GATE Institute at Sofia University “St. Kliment Ohridski,” Eastern Europe’s first Center of Excellence in Big Data and Artificial Intelligence. The Institute contributed expertise to two working groups responsible for drafting the Code of Conduct: one focused on “Transparency and Copyright” and the other on “Risk Management in Organizations.”

Ivo Emanuilov, Head of the Experimental Regulation and Digital Policies Laboratory at GATE and a participant in these groups, emphasized the importance of translating ethical and legal principles into practical technology governance. He highlighted the Institute’s dedication to fostering a secure and responsible AI ecosystem while providing a space to test and refine policies in real conditions.

To further its mission, GATE recently launched the Experimental Regulation and Digital Policy Lab. This innovative facility blends legal, ethical, and technological perspectives to prototype regulatory frameworks and test AI solutions in controlled environments before market release. The Lab is Bulgaria’s first of its kind and among a few in the EU offering such hands-on evaluation of AI’s societal impact and safety.

Through these efforts, the GATE Institute is helping bridge the gap between policy-making and real-world application, supporting the European Union’s aim to ensure AI technologies develop in ways that respect human rights and foster public trust.

Source: press release