Data transparency and copyright: opportunities and challenges

Transparency is widely recognised as a core principle of AI governance, yet its practical implementation remains challenging. The data transparency obligations in the EU AI Act, introduced to support copyright enforcement, illustrate the difficulty of designing transparency rules that are effective and capable of fairly balancing the interests of right holders and AI developers. This article argues that these measures are unlikely to achieve their intended objectives. In particular, the dataset summaries required by Article 53(1)(d) of the AI Act may provide useful information to creative sectors but are unlikely to enable individual right holders to identify infringements or effectively enforce their rights. As a result, the EU approach risks imposing significant compliance costs without delivering meaningful copyright protection. The article argues that other jurisdictions should carefully evaluate the EU model before adopting similar measures. Instead, transparency should form part of a broader reform of copyright law encompassing rights, exceptions, licensing mechanisms, and liability rules to achieve a more balanced framework for AI and copyright.