This paper presents a conceptual framework that fuses corpus-based analysis (distant reading) with generative AI practice (distant writing) to advance critical digital literacy in secondary and tertiary classrooms. Using accessible corpus tools—COCA, DWDS, and Google Ngram Viewer—together with large language models such as ChatGPT, the framework enables students to interrogate large-scale linguistic patterns, generate context-specific texts, and assess algorithmic bias. A classroom scenario on German parliamentary discourse about migration illustrates the two-step sequence: learners first trace diachronic shifts in key terms (e.g., Gastarbeiter, Migranten, Asylanten) and their collocational profiles; they then prompt an LLM to draft historically anchored speeches, revising its output against corpus evidence. Through this iterative comparison, students leverage corpus skills (selecting, querying, interpreting corpora) to critically engage with LLMs and develop AI literacy (understanding model training, detecting confabulations, refining prompts). Combining these literacies foregrounds the epistemological stakes of data selection, metadata transparency, and sociopolitical bias, equipping learners to navigate and critique the contemporary textual landscape.