On August 2, updated transparency rules for general purpose artificial intelligence systems entered into force in the European Union. About this informs DW.
Innovations concern such AI systems as ChatGPT and Gemini, which are able to generate texts, images or perform software tasks. From now on, the developers of such models are obliged to publicly disclose information about the principles of AI operation and data sources on which training was carried out.
In the case of the most powerful models that may pose risks to society, it will be mandatory to document the security measures implemented. In addition, the new rules aim to protect the rights of authors and content owners, including intellectual property. From August 2, natural persons will be able to file lawsuits against AI developers in case of violations.
At the European level, monitoring of compliance with these requirements will be carried out by a specially created body — the European AI Office. Inspections for new models will begin in August 2026, and for models launched before August 2, 2025, in August 2027. According to experts, potential fines for violations could reach up to $17.1 million, or 3% of the annual volume of the AI services market.
We will remind you that in March 2024, the European Parliament approved the Artificial Intelligence Act, which introduces systematic regulation of this industry.
