NETHERLANDS Trends and Developments Contributed by: Radboud Ribbert and Wouter van Wengen, Greenberg Traurig, LLP
collective rights organisations’ websites to see which works include the opt-out criteria; • include the necessary contractual protec - tions (two types of agreements are of particu - lar relevance in terms of machine learning, namely (i) agreements with the owner of the dataset on which the AI model is trained and (ii) agreements with the customer of the AI model; in both cases, consider creating an agreement with a fair and balanced distribu - tion of liability regarding inadvertent use of opted-out copyright-protected works); and • for copyright-protected content that may be used for TDM purposes, consider implement - ing technical and organisational limitations on the use of the content to ensure it is only used for training the AI model. Regulating Deepfakes as Copyright or Neighbouring Right Deepfakes are a specific use of artificial intel - ligence, whereby the technology is used to cre - ate highly realistic AI-generated audio, video, or images. The use of deepfakes creates issues for rights-holders of copyrights and neighbouring rights of the original works that the deepfakes are based on. For instance, deepfakes frequently rely on content protected by copyrights or neigh - bouring rights as training data or as a compo - nent of the generated output. This use sparks a regulatory debate regarding whether such use is permitted without the rights-holders’ consent. In addition, deepfakes give rise to other questions, such as who should be considered the owner of the resulting content and whether the deep - fakes lead to misappropriation of identity when the deepfakes replicate the likeness of individu - als for commercial use. In addition to the opt-out rights discussed above, rights-holders may soon have an alternative way under Dutch laws to protect their position when
companies create deepfakes by means of a pro - posal for a new law. The Dutch Secretary of Justice’s note, pub - lished on 15 October 2024, details the proposed Strengthening Authors’ Contract Law (ACR 2). This bill builds on the 2015 ACR law, enhanc - ing authors’ bargaining power and addressing new challenges, such as the protection of heirs and the threat of deepfakes. It introduces collec - tive bargaining power for authors, supported by industry platforms, and responds to AI’s impact on copyright. The proposal is expected to gain strong parliamentary support. Further updates are planned five years after implementation. Expansion of Permitted Referential Use Coming to a Halt Under EU law, referential use of a trade mark is permitted under certain conditions, particularly when it is used to identify or refer to the goods or services covered by the trade mark. The standard for applying this exception was previ - ously very high. The referential use had to be deemed necessary, meaning it was the only way to provide the public with clear and complete information about the purpose of the underlying product, in order to preserve free competition in the market for that product. However, in the recent Court of Justice EU deci - sion in Inditex v Buongiorno (11 January 2024), the threshold for this necessity was significantly lowered. The CJEU ruled that necessity is just one reason for allowing referential use, without establishing clear boundaries for what qualifies as permitted referential use. The trend in recent court decisions appears to favour referential use. A notable example in the Netherlands is the Dutch Supreme Court ruling in the Jiskefet case, involving a popular Dutch comedy group.
375 CHAMBERS.COM
Powered by FlippingBook