Detecting RAG Advertisements Across Advertising Styles

TL;DR

This paper develops a taxonomy of advertising styles for LLM-generated ads, evaluates detection methods' robustness to style changes, and highlights the need for efficient, resilient detection models.

Contribution

It introduces a new taxonomy of advertising styles, simulates evasion tactics, and assesses detection approaches, emphasizing the importance of lightweight, robust models.

Findings

Entity recognition models are highly effective at detecting ads.

Detection models are largely robust to style changes.

Lightweight models like random forests are brittle under style variations.

Abstract

Large language models (LLMs) enable a new form of advertising for retrieval-augmented generation (RAG) systems in which organic responses are blended with contextually relevant ads. The prospect of such "generated native ads" has sparked interest in whether they can be detected automatically. Existing datasets, however, do not reflect the diversity of advertising styles discussed in the marketing literature. In this paper, we (1) develop a taxonomy of advertising styles for LLMs, combining the style dimensions of explicitness and type of appeal, (2) simulate that advertisers may attempt to evade detection by changing their advertising style, and (3) evaluate a variety of ad-detection approaches with respect to their robustness under these changes. Expanding previous work on ad detection, we train models that use entity recognition to exactly locate an ad in an LLM response and find them to be both very effective at detecting responses with ads and largely robust to changes in the advertising style. Since ad blocking will be performed on low-resource end-user devices, we include lightweight models like random forests and SVMs in our evaluation. These models, however, are brittle under such changes, highlighting the need for further efficiency-oriented research for a practical approach to blocking of generated ads.