This study investigates the application of generative artificial intelligence in creative generation and style consistency modeling for visual advertising. By analyzing semantic control mechanisms, image style feature extraction, and consistency modeling techniques in ad content generation, a systematic generative framework for automated advertising design is outlined. The proposed approach leverages prompt-driven input, multimodal integration, and style vector constraints to construct a collaborative system combining prompt parsing with style control modules. Results indicate that the model enables rapid and diversified advertising output while maintaining brand tonality and visual coherence, confirming the feasibility and adaptability of generative AI in commercial image creation. The findings provide methodological and system-level references for intelligent design and stylistic regulation in visual advertising.