With the growing integration of artificial intelligence in education, teachers' ability to use these tools in designing intelligent, interactive learning activities has become increasingly important. This is particularly significant in special needs education, where flexible instructional design plays a crucial role. Effective use of such technological capabilities largely depends on teachers' AI literacy. Accordingly, this study examined the relationship between the four components of AI literacy among special education teachers—cognitive, behavioral, affective, and ethical—and their use of AI capabilities in designing instructional activities. This study employed a descriptive–correlational survey design. The statistical population consisted of special education teachers in Alborz Province, Iran, from which 142 teachers were selected through convenience sampling. Data were collected using the AI Literacy Questionnaire by Ng et al. and a researcher-developed checklist measuring teachers' use of AI capabilities in instructional design. Data were analyzed using Pearson correlation and multiple regression. Findings revealed a positive, significant relationship between teachers' AI literacy and their use of AI capabilities in instructional design (r = 0.522, p < 0.01). Regression analysis further indicated that cognitive, behavioral, and ethical components significantly predicted teachers' use of AI capabilities, whereas the affective component was not a significant predictor. Results highlight the importance of professional development programs that go beyond technical skills and strengthen conceptual understanding, practical competencies, and ethical awareness in educational AI use.