Researchers have introduced SlopShape, a novel methodology designed to identify AI-generated commercial web content by analyzing structural signatures rather than word-level patterns. Moving beyond brittle linguistic detectors, this approach examines how information is organized, the sequence of presentation, and the specific use of evidence. The researchers replicated findings from StoryScope by comparing 2,250 human-authored blog posts from 268 company domains against 11,250 AI-generated mirrors produced by five frontier models. Using a 214-feature instrument applied by a Large Language Model—which achieved high validation scores in human-human (0.928 kappa) and human-model (0.946 kappa) annotation sessions—the system identified AI posts using only 187 structural features. The model achieved a 98.0 macro-F1 score on held-out companies, a performance that remained stable at 98.1 even when AI posts underwent model-based rewording. SlopShape successfully attributes 79.3% of posts to their correct source model, significantly outperforming the 16.7% chance rate. The study concludes that AI-generated content exhibits a distinct, "self-announcing" structural shape, whereas human-authored posts occupy rare and unique configurations. The authors have released the complete pipeline, including prompts and code, to facilitate further research into structural text attribution.