The proliferation of sophisticated art forgeries poses mounting challenges for authentication practices in today's art market. This paper introduces a novel framework that leverages generative artificial intelligence to verify artwork authenticity by analyzing artistic style consistency. We conceptualize artist-specific stylistic signatures as "style genes" and employ fine-tuned diffusion models to extract and analyze these inherent characteristics. This approach combines prompt engineering techniques with inverse style-matching protocols to assess whether questioned artworks align with the stylistic fingerprints of attributed artists. Experimental validation on master Chinese ink paintings (including Qi Baishi, Xu Beihong, and Zhang Daqian) and Western oil paintings (including Picasso, Monet, and Van Gogh) demonstrates superior performance compared to traditional methods, achieving 94.3% accuracy in forgery detection while maintaining interpretability through multimodal contextual analysis (integrating visual, textual, and historical data).
