This paper examines how Generative Artificial Intelligence (GenAI) in K-12 education interacts with teachers' Pedagogical Content Knowledge (PCK). Using the Refined Consensus Model (RCM) as an analytical framework, we analysed how GenAI influences the three PCK domains: personal PCK (pPCK), collective PCK (cPCK), and enacted PCK (ePCK). Guided by the research question "How can teachers and GenAI collaborate to develop and use PCK and pedagogical reasoning in K-12?", we analysed three researcher-constructed vignettes – hypothetical but plausible scenarios representing GenAI use in planning, classroom interaction, and assessment – to derive analytical propositions about the PCK-GenAI relationship. Our analysis suggests tensions between traditional PCK boundaries and GenAI-mediated practices, particularly regarding the RCM's plan-teach-reflect cycle. We propose that GenAI can amplify teachers' PCK when critically evaluated, but poses risks for novice teachers whose pPCK is still developing. The paper identifies how GenAI reconfigures boundaries between personal and collective knowledge, reshapes enacted practice when AI becomes a co-constructor of teaching, and highlights the continued centrality of teacher expertise. We propose directions for supporting effective human-GenAI collaboration in PCK development and discuss implications for teacher education.