NETHERLANDS Trends and Developments Contributed by: Radboud Ribbert and Wouter van Wengen, Greenberg Traurig, LLP
Introduction The interplay between innovation and intellec - tual property (IP) rights is becoming increasingly complex as new technologies reshape indus - tries. From AI training data to the boundaries of referential trade mark use, recent developments highlight the challenges of balancing creativity, commerce, and legal protections. Key areas of focus include the impact of the EU Artificial Intelligence Act (EU AI Act) on copyright compliance in AI model training, recent rulings on the permissible scope of trade mark use, and expanded copyright protections for international applied art under EU law. These changes illus - trate the evolving legal landscape and its impli - cations for creators, companies, and consum - ers. This discussion examines these issues, offer - ing insights into the regulations and decisions shaping the future of IP in a rapidly advancing technological era. Using Existing Works for Training Your AI Model AI’s process of learning mirrors human knowl - edge acquisition, such as reading a book or lis - tening to music, which are generally unregulated if there is no direct copying. However, unlike humans, AI training is controlled and involves processing vast amounts of data, potentially accelerating innovation but raising concerns about intellectual property (IP) rights. The criti - cal regulatory challenge is determining how far machine learning should be restricted to uphold copyright protections. With the entry into force of the EU AI Act, com - panies involved in training AI models need to carefully evaluate the datasets they use, as cer - tain copyright-protected works may no longer
be available for this purpose. The EU AI Act contains a provision that equates AI/machine leaning with “text and data mining” (TDM) under the EU Text and Mining Directive. Consequently, “machine learning” is allowed, provided that the developer of the machine learning functionality had “lawful access” to the content “for the pur - pose of text and data extraction” and the owner of the respective copyrights in the content did not expressly reserve the extraction of text and data (ie, the “opt-out mechanism”). Although the process for submitting valid opt- out requests is unclear, organisations like the Dutch organisation Pictoright (visual art – see also below) and the French organisation Sacem (music) have developed rights reservation state - ments to prevent their works from being used in AI training. Similar opt-out notices are also appearing on websites and social media con - tent. While no definitive legal precedents exist to val - idate these opt-out statements, their usage is expected to increase following the EU AI Act’s implementation. This regulation highlights the growing effort to balance AI innovation with the protection of intellectual property rights. AI system providers and developers should con - sider implementing measures and configurations to avoid infringement claims by rights-holders. Below are four additional considerations: • in the process of web scraping or review - ing pre-built datasets, confirm whether the content used for machine learning purposes is subject to access restrictions, such as a paywall or other (technical) restrictions; • consider verifying that the rights-holders have not reserved the right to make reproductions for TDM purposes by, for example, searching
374 CHAMBERS.COM
Powered by FlippingBook