The European Union has introduced a new AI code that urges companies to disclose their training data and avoid copyright violations. This code is part of the EU's broader effort to regulate AI development and ensure transparency, safety, and respect for fundamental human rights. Under the new code, companies must document and publicly disclose comprehensive summaries of their training data, including data sources and processing methods.
The code also emphasizes the importance of copyright compliance, requiring providers of general-purpose AI models to implement measures to respect copyright and related rights. This includes identifying and removing content for which rights have been reserved. Additionally, the code outlines concrete practices for managing systemic risks associated with advanced AI systems, including potential threats to citizens' personal safety and rights.
Companies that fail to comply with the AI Act may face significant fines, with penalties reaching up to 7% of their annual sales or 3% for companies developing advanced AI models. While the code is voluntary, signing on will provide companies with increased legal certainty and lower administrative burdens. Existing AI models have a two-year grace period to comply, while new models must adhere to the regulations within one year.
The introduction of this code has sparked debate among industry leaders, with some tech giants expressing dissatisfaction with the requirements. Companies like Meta and Alphabet have cited concerns about revealing sensitive information and the potential impact on innovation. Meanwhile, some European companies have requested a suspension of the AI Act's implementation, calling for a more "innovation-friendly regulatory approach". Despite these concerns, the EU's efforts to regulate AI development aim to promote transparency, safety, and respect for human rights in the rapidly evolving field of artificial intelligence.
