Digital health literacy (DHL) is central to supporting engagement with increasingly digitalised healthcare systems. However, dominant models remain grounded in a linear, skills-based logic, focused on access and use, that assumes relatively stable and transparent information environments. This conceptual position paper asks: How adequate are current models of digital health literacy in AI-mediated healthcare contexts? We define AI-mediated care as environments in which interactions, decisions, and care pathways are increasingly shaped by algorithmic systems that analyse, prioritise, and generate recommendations, repositioning individuals as participants in automated decision-making processes. In response, we examine two key questions: (1) What new forms of user capability are required in algorithmically mediated systems? and (2) How can digital health literacy be reconceptualised to reflect these changing conditions? We argue that existing frameworks are no longer sufficient as healthcare shifts toward platform-based, data-driven infrastructures. Users are not simply “using” technologies but navigating complex, adaptive systems. We employ a conceptual, critical synthesis of existing literature to develop and advance theoretical frameworks. We propose a reconceptualisation of DHL as system navigation capability, advancing a model comprising interpretation, questioning, navigating, and negotiating. This framework captures the critical competencies required to engage with dynamic systems, including interpreting outputs, recognising bias, and negotiating decisions within partially automated care pathways. A simple conceptual model illustrates this shift from functional engagement to relational and systemic interaction. Without such reframing, digital inclusion efforts risk falling behind, particularly for midlife and older adults, who remain vulnerable to emerging forms of exclusion in AI-mediated healthcare environments.