Summary:
The draft of the Code of Practice on General-Purpose Artificial Intelligence (GPAI) currently under consideration by the European Commission aims to establish rigorous legal frameworks for providers of AI models, including those such as ChatGPT, Google Gemini, and Midjourney. The draft introduces significant changes, particularly in addressing the issue of authorship and legends, with new provisions such as increased transparency requirements. However, the draft also faces critical challenges, particularly concerning copyright law, which violates fundamental rights to know and to remain silent, especially if companies rely on AI models without authorization. Industry representatives, along with lobby groups like the CCP and the Silicon withdraw their support, highlighting concerns about the draft’s alignment with existing EU laws and its potential to infringe upon intellectual property rights. The draft has also failed to meet some of the key obligations of the AI Act, leaving industry operators and companies on auttitude to deal with complexities in controlling AI technologies.
The draft emphasizes the need for risk assessment measures, but its limitations suggest there may be deeper issues related to the duty to protect fundamental rights. While the draft acknowledges the importance of protecting intellectual property, it also flags the need for more comprehensive ethical practices and risk assessments. The legislative process is currently delayed, with a deadline set for submissions on March 30, pending comments and reviewer input. However, if the draft is finalised, it could influence ongoing discussions about the role of universities and research institutions in ensuring that AI technologies are developed responsibly. The draft also comes at a time when copyright law in the EU continues to evolve, with successor legislation still in press. The potential for copyright disputes with AI providers could have far-reaching implications for the commercialization and regulation of AI technologies in Europe.