While generative AI has enhanced digital multimodal composing (DMC) processes, growing concerns about learners’ overreliance on AI highlight the need to foster critical AI literacy. This study first designed an instructional model based on theoretically informed components of critical AI literacy and then implemented it to explore how learners’ AI use and perceptions evolved during AI-assisted DMC. Data were collected from two high school-aged participants, including screen recordings of their DMC practices before and after the intervention, stimulated recalls, and semi-structured interviews. Findings show that both participants developed more critical use of AI tools, such as cross-validation of outputs, strategic prompting, and more independent composing. Their perceptions also reflected a more balanced redistribution of agency and authorship, as they selectively used AI tools to cope with technical and linguistic challenges. The study suggests that critical AI literacy can be meaningfully developed when process-based and critical AI literacy-informed instructional intervention is provided.

1. INTRODUCTION
2. REVIEW OF THE LITERATURE
    2.1. Multiliteracies and DMC
    2.2. DMC and L2 Writing Development
    2.3. Challenges in AI-Assisted DMC
3. THE ABCDE MODEL: INSTRUCTIONAL FRAMEWORK FORCRITICAL AI LITERACY IN DMC
    3.1. Theoretical Framework: Critical AI Literacy
    3.2. Exemplary Activities and Materials
4. CASE STUDY
    4.1. Participants
    4.2. Tasks and Tools
    4.3. Instructional Intervention
    4.4. Data Collection and Analysis
    4.5. Findings
5. DISCUSSION
6. CONCLUSION
REFERENCES
APPENDIX

• Dong-Yeob Ham(Graduate Student, Department of English Language Education, Seoul National University)
• In Chull Jang(Professor, Department of English Language Education, Seoul National University; 1 Gwanak-ro, Gwanak-gu, Seoul 08826, Korea) Corresponding author