<p>The present article puts forward a critical reinterpretation of the theory of journalism in the context of artificial intelligence (AI). The article proposes that AI should not be conceived of as an instrumental innovation, but rather as a socio-technical regime that reconfigures journalistic action, information mediation, and the production of public knowledge. The model is developed through a critical integrative review of classical communication theory and contemporary scholarship on AI, automation, platformisation and datafication in journalism. On this basis, it proposes the Integrated Sociotechnical Model of Journalistic Action in AI (ISMJA-AI), which conceptualises journalism as a system of agency distributed among journalists, algorithmic infrastructures, platforms and audiences. The model enables the analysis of non-linear dynamics of information production, circulation and validation, as well as structural tensions between professional autonomy, automation, pluralism and algorithmic governance. The article posits that contemporary journalism functions as an augmented epistemic institution, whose democratic efficacy is contingent on its capacity to critically engage with opaque AI systems. The ISMJA-AI is presented as a theoretical framework for research, empirical analysis and normative debate on journalism, artificial intelligence and democracy.</p>