Mobile advertisements (ads) fuel the app economy, yet poorly regulated ads can degrade user experience and expose users to malware. Identifying ad regulation violations is challenging because ads are dynamically retrieved from remote servers and embedded across various user interfaces (UIs), making them difficult to locate and trigger. To address these challenges, we propose AdWise, which harnesses Large Language Models (LLMs) to unify static ad analysis and dynamic UI exploration for detecting mobile ads and assisting ad regulation. AdWise first develops novel static analysis to identify UI widgets used to place ads (ad widgets). To reach these widgets, AdWise instructs an LLM to guide dynamic exploration using three types of guidance derived from static analysis and ad placement practices: (1) WTG-based guidance, which uses a statically built window transition graph (WTG) to delineate the app's structural topology; (2) semantic guidance, which reasons about app functionality to prioritize paths users are likely to visit; and (3) structural guidance, which uses retrieval-augmented generation (RAG) to recognize common ad-heavy UI layouts from a knowledge base. Our evaluation on 100 benchmark apps shows that AdWise outperforms the strongest baseline by 25.83 percentage points in ad-widget detection and uncovers 34.34% more violations across six violation types in three regulation categories. A broader in-the-wild study identified 383 regulation violations within 2,047 apps using Meta Ads. We reported the findings to Google Play and the Abuse Track of the Meta Bug Bounty program.

翻译：暂无翻译