This study investigated how a structured generative artificial intelligence (GenAI)-based professional development program, which builds instructional practices of teachers, is effective. The research used a quasi-experimental pretest-posttest control-group design and 80 teachers of public schools in Ma’an Governorate, Jordan were used. There were equal numbers of participants in both experimental (n = 40) and control (n = 40) groups. The intervention was eight training sessions conducted in four weeks and was based on prompt engineering, lesson planning, instructional design, differentiated learning, assessment construction, feedback generation, verification of AI outputs, and ethical and privacy considerations. A 24- item Generative AI-Enhanced Instructional Practices Scale was used to collect data, which included three dimensions: lesson planning, instructional design and teaching practices, and assessment and feedback practices. The analysis of covariance (ANCOVA) and Pearson correlation coefficients, as well as descriptive statistics, Cronbach alpha, independent-samples t tests, and analysis of covariance (ANCOVA) were performed using IBM SPSS. The findings showed acceptable-high internal consistency, baseline group equivalence, and statistically significant adjusted posttest differences in favor of the experimental group in all dimensions. Big effect sizes were realized. The results suggest that an instructional GenAI training program based on pedagogy can enhance lesson planning, instructional design, assessment, and feedback practices of teachers. The research adds a systematic intervention framework and empirical data of the public schools in Ma’an Governorate.