This qualitative action research study examines how the implementation of a pedagogical intervention using Generative Artificial Intelligence (GenAI) tools influenced eighth-grade students' writing practices and their understanding of responsible AI use in English as a second language. Conducted at a private, bilingual, all-girls school in Bogotá, Colombia, the study involved five structured interventions in which GenAI tools were deliberately restricted to the editing stage of the writing process, preserving students' authorial ownership across three writing products: reading diary paragraphs, horror flash fiction stories, and poems. Data were collected through classroom recordings, student writing artifacts, teacher and peer feedback, a reflective diary, focus groups, and a thinking routine, and analyzed through thematic coding. Findings reveal three interconnected outcomes: substantive improvements in students' written productions across sentence fluency, word choice, and literary expression; the emergence of metacognitive awareness through socially mediated, multi-source feedback processes; and the development of critical AI literacy, including the capacity to evaluate, resist, and selectively appropriate AI-generated suggestions. The study concludes that effective GenAI integration in L2 writing instruction requires explicit scaffolding, ethical grounding, and sustained teacher mediation.