The European Commission has taken a significant step in regulating artificial intelligence by appointing a 13-member team tasked with drafting a code of practice for general purpose AI, as mandated by the recently enacted AI Act. This initiative is designed to address various aspects of AI governance, ensuring that the development and deployment of AI technologies are conducted responsibly and transparently. The announcement, made on Monday, outlined the formation of four working groups that will focus on crucial areas such as transparency, copyrights, risk assessment, and governance related to AI systems, particularly large language models. The code of practice is expected to come into effect next August.
Each working group comprises chairs and vice-chairs with expertise spanning computer science, AI governance, and law, reflecting a comprehensive approach to developing these regulations. The European Commission emphasized the importance of this leadership in refining and drafting the rules. For instance, the co-chairs of the transparency and copyright working group combine backgrounds in European copyright law and AI transparency, ensuring a well-rounded perspective in creating guidelines that balance innovation with legal and ethical considerations.
The AI Act, which came into force in August, includes specific mandates requiring developers to disclose copyrighted materials utilized in training AI models. Companies that fail to comply with this new legislation could face severe penalties, including fines reaching up to 35 million euros or 7% of their annual turnover. This regulatory framework aims to promote accountability and transparency within the AI sector, holding developers responsible for their AI systems’ implications on society.
In the drafting process, the EU AI Office has engaged nearly 1,000 stakeholders in discussions to gather diverse input. Following an initial meeting in July, the commission received over 450 responses, highlighting significant interest and investment in shaping these new rules. The EU AI Office is also planning a workshop for general purpose AI developers in mid-October, further facilitating collaboration and dialogue among industry players. Notably, just days before this announcement, over 100 companies had signed a voluntary AI pact, though major players like Meta, Apple, and MistralAI have chosen not to participate, indicating varied approaches to AI governance within the industry.
