An Effective Framework for Cross-Site Scripting Payload Generation

In the field of cybersecurity, machine learning techniques are being increasingly utilized for cross-site scripting (XSS) vulnerability detection. These techniques playa crucial role in facilitating the identification process and reducing the heavy reliance on human analytics. This paper puts forward a novel approach for generating the payload of cross-site attack scripts. Initially, datasets are gathered for model training, and positional coding is incorporated into the input data to enhance the model's understanding of the structure. The generative AI model conducts an in-depth analysis of the dataset and then further combines the syntactic features of cross-site scripting with common bypass methods to formulate adaptive mutation strategies. Based on the attack response results, the model parameters are updated, and the payload generation process is iteratively optimized. The experimental results demonstrate that this method can generate highly deceptive malicious scripts with a notably high success rate.