EU Introduces Comprehensive AI Code with Focus on Copyright and Transparency
The EU has launched a new code to support compliance with the AI Act, focusing on copyright protections and transparency requirements.
Key Points
- • The EU's AI code includes copyright protections and transparency mandates for AI systems.
- • Developers must maintain documentation of their AI's features for regulators.
- • Companies are prohibited from training AI on pirated materials and must address copyright violations.
- • The code is voluntary, yet it imposes fines up to 7% of annual sales for breaches.
On July 10, 2025, the European Union unveiled a new code of practice that outlines essential guidelines for compliance with the forthcoming AI Act, particularly emphasizing copyright protections and transparency requirements for artificial intelligence systems. This rollout aims to assist companies in navigating the complexities of the new regulatory landscape.
The code mandates that developers provide thorough documentation regarding the functionalities of their AI models to regulators and third parties. It also prohibits the training of AI on pirated material and necessitates adherence to copyright requests from content creators. Companies are now expected to implement procedures to remedy any instances where AI-generated content infringes on copyright laws. Violations of the AI Act can result in severe penalties, potentially amounting to 7% of a company's annual revenue, while firms engaged in the development of advanced AI technologies face fines of up to 3% for breaches.
Although the code is intended to be voluntary, its impact is significant as it offers a framework for businesses aiming to establish compliance with the AI Act effectively. Notably, the code will be applicable in phases, with specific regulations for high-risk AI technologies, such as general-purpose systems like ChatGPT and Claude, set to be enforced next month.
The announcement has not been without controversy. Major technology corporations, including Meta and Alphabet, have expressed concerns that the code exceeds the intended scope of the AI Act and may inhibit innovation. A consortium of European firms, including influential names like ASML and Airbus, has called for a two-year delay in implementing the AI Act to foster a more innovation-friendly regulatory environment. Despite these calls, the European Commission, which has faced criticism for delayed publication of the code, has refused to postpone the rollout, emphasizing the involvement of multiple stakeholders in its development.
As it stands, enforcement of the AI Act will not commence until August 2026, during which time national courts will oversee compliance, although there may be challenges related to their technical expertise. The ongoing discussions highlight the tension between regulatory oversight and the desire to nurture a thriving AI industry within Europe.