The integration of generative Artificial Intelligence (AI) technologies into education is rapidly transforming how knowledge is produced, distributed, and reused. As AI tools grow more sophisticated and accessible, educators and students are increasingly employing them to generate teaching materials, including textual, visual, and multimedia content. However, this widespread adoption introduces considerable legal uncertainty regarding the attribution, recognition, and management of intellectual property rights (IPR) over such AI-assisted outputs. This paper investigates these challenges from a legal-technical perspective, focusing on how the concepts of authorship, ownership, and control are reshaped in educational contexts by the presence of generative AI.

The methodological approach is rooted in comparative legal analysis within the European framework, drawing upon statutory law, national doctrines, institutional policies, and relevant jurisprudence. The study analyzes two principal models of authorship: the anthropocentric model, which confines authorship to human agents providing creative and original input; and the hybrid model, which considers AI a substantial contributor in a co-authorship regime. Both models present interpretative and operational challenges, particularly with respect to the legal notion of originality, the allocation of moral and economic rights, and the attribution of liability for generated content.

In this evolving legal scenario, opt-out clauses are emerging as an increasingly adopted contractual mechanism. These clauses permit human creators to prevent the inclusion of their content in AI training datasets and related processing. They are grounded in legal principles such as autonomy, informed consent, and the protection of proprietary data. Their function is to give authors the ability to control downstream uses of their materials, particularly where reuse for algorithmic purposes may conflict with the author’s original intent or legal rights.

Nonetheless, the use of opt-out clauses is not without complexity. Their legal effectiveness depends on multiple factors: the clarity of terms and mechanisms of consent; the consistency of enforcement across diverse legal jurisdictions; and the retroactive applicability of such exclusions, particularly the feasibility of removing previously collected data from AI models. In addition, determining the authorship of hybrid content remains problematic, especially when the human role is minimal or mediated by opaque algorithmic processes. The paper also explores the risk that excessive reliance on opt-outs could reduce the availability of diverse, high-quality training data for educational AI, potentially slowing technological innovation and widening equity gaps between institutions.

The findings emphasize that while opt-out clauses can play a critical role in reinforcing the agency of content creators, they must be supported by broader regulatory and institutional measures. A coherent legal response should include harmonization of copyright standards, transparency obligations for AI developers, and transnational coordination mechanisms. As educational institutions increasingly incorporate AI into learning environments, an anticipatory legal framework is necessary—one capable of safeguarding intellectual property without inhibiting the responsible and inclusive development of generative technologies.

Keywords: Intellectual Property, Education, Law, Ethic.