New EU Regulations Demand Transparency from AI Model Providers

The European Union has implemented new regulations that require artificial intelligence (AI) model providers to disclose vital information about their training data and operational mechanisms. Starting from August 2, 2025, companies developing general-purpose AI systems, such as ChatGPT and Gemini, must adhere to these transparency obligations. These systems are capable of performing a variety of functions, including text generation, speech analysis, and programming tasks.

According to the new guidelines, AI operators are mandated to reveal the nature of their models and the data sources utilized for training. This initiative is rooted in the EU's AI Act, which was enacted in May 2024 and aims to enhance safety and accountability within the AI sector. Notably, especially advanced AI models, which may pose risks to society, are required to implement and document necessary safety measures.

Moreover, these regulations are designed to reinforce copyright protections. Developers must now disclose the sources of their training data and clarify whether they have utilized automatic data scraping from websites. They are also tasked with outlining the measures taken to safeguard intellectual property rights. Additionally, the EU stipulates that companies must establish a point of contact for rights holders.

Despite these advancements, several national and international collectives of authors, artists, and publishers have expressed concerns regarding the adequacy of the legislation in protecting intellectual property. The Initiative for Copyright has criticized the lack of a requirement for disclosing specific datasets, domains, or sources, arguing that the current measures may be insufficient.

The enforcement of these regulations will be overseen by the newly established European Office for Artificial Intelligence, which is set to begin monitoring new AI models from August 2026. For those models released before August 2, 2025, the office will commence oversight in August 2027. Companies found in violation of these regulations could face fines of up to 15 million euros or three percent of their total global annual revenue.

As the landscape of artificial intelligence continues to evolve, these new EU regulations mark a significant step toward ensuring accountability and transparency in AI development and deployment. The implications of these requirements will undoubtedly influence how AI technologies are created and utilized in the future.