Abstract
The contemporary information environment is marked by the increasing ease with which journalistic stories can be generated, reformatted, and adapted through generative AI systems. As a result, content becomes increasingly liquid, and the article no longer functions as the sole or stable unit of journalism. While writing remains central to journalistic practice, the primary site of value shifts toward the production and stewardship of reliable information from which multiple narratives can be formed. This paper argues that, in AI-mediated news and journalism, journalists increasingly operate not only as authors of finished stories but as producers and curators of reusable information artefacts that support both human and AI-assisted storytelling. Situated within discussions of AI-mediated news, liquid content, modular journalism, and data storytelling, the paper advances the need for a modular and compositional storytelling structure suited to dynamically generated narratives. Building on prior empirical work, it positions the Water Tower storytelling model as a structure capable of organising liquid journalistic content into coherent stories across formats and levels of detail, while remaining grounded in editorial judgement.