Beyond Possessive Agency: TikTok, YouTube, and the Inadequacies of GDPR, OSA, DSA, and AIA

This paper critiques the foundational assumption underpinning UK and EU regulations—including the GDPR, Online Safety Act, Digital Services Act, and Artificial Intelligence Act—that agency is a possessive attribute rooted in individual autonomy. Using YouTube and TikTok as case studies, it examines how advanced recommendation systems powered by deep neural networks, multi-armed bandits, and reinforcement learning blur the boundaries between user agency and platform influence. Drawing on feminist relational theory and Karen Barad’s concepts of intra-action and diffraction, the paper argues that contemporary platforms generate a distinct, relational form of agency that operates interstitially in the ‘in-between’ of user actions and algorithmic systems. This emergent directive power challenges existing law in the EU and UK concerning online safety, which position platforms as neutral tools rather than co-constitutive entities. The paper calls for a re-imagination of legal ontologies and for a shift from possessive agency to relational governance to address the complexities and risks posed by modern algorithmic assemblages.