This conceptual article examines the role of "speculation" in driving responses to generative AI platforms in literacy education and the implications for research, pedagogy, and practice. Our focus on "speculation" encompasses two meanings of the term -- each of which has inspired lively lines of inquiry in literacy studies and transdisciplinary research on artificial intelligence, respectively. In the first sense, literacy scholars have recognized literacy education as a speculative project -- one characterized by the cultivation of particular reading and writing practices in order to prefigure different imagined social futures. In the second sense, scholars of media and computational cultural studies have theorized a different kind of speculative logic that underwrites the design and functioning of AI platforms -- one characterized by extrapolative prediction and algorithmic reasoning. Investigating the evolving relationship between these modes of speculation, we argue that the former has allowed literacy education to be uniquely susceptible to the influence of the latter; and likewise, that the latter exerts its influence in ways that remake the former in its image. We theorize this relation as a process of "speculative capture," and we highlight its stakes for equitable literacy education. We then conclude by providing provocations for researchers and teachers that may be of use in preempting the collapse of these speculative formations into one another; and perhaps, in mobilizing a conception of the speculative that works productively toward alternative ethico-political ends.