Detecting RAG Advertisements Across Advertising Styles

This project addresses a nascent but critical problem: the 'ad-supported LLM.' As platforms like Perplexity, SearchGPT, and Bing evolve, the blending of organic RAG content with paid promotion is inevitable. The project's strength lies in its academic foundation—specifically a taxonomy that bridges marketing literature with LLM outputs. However, with 0 stars and only 7 forks, it currently lacks the community momentum or data gravity required for a high defensibility score. Its current state is a research artifact rather than a production tool. The 'moat' here is purely intellectual/methodological, which is easily replicated once published. The platform domination risk is high because the frontier labs (OpenAI, Google) control the inference pipeline and could easily include headers or metadata that render third-party detection redundant, or conversely, they could employ adversarial techniques to make ads indistinguishable from organic content. The displacement horizon is 1-2 years, coinciding with the expected rollout of large-scale monetization in RAG-based search engines. Competitors would include browser-based ad-blockers (e.g., uBlock Origin) and LLM-guardrail companies (e.g., Arthur, Lakera) moving into the 'transparency' space.