CV-Inspector: Towards Automating Detection of Adblock Circumvention

Abstract

The adblocking arms race has escalated over the last few years. An entire new ecosystem of circumvention (CV) services has recently emerged that aims to bypass adblockers by obfuscating site content, making it difficult for adblocking filter lists to distinguish between ads and functional content. In this paper, we investigate recent anti-circumvention efforts by the adblocking community that leverage custom filter lists. In particular, we analyze the anti-circumvention filter list (ACVL), which supports advanced filter rules with enriched syntax and capabilities designed specifically to counter circumvention. We show that keeping ACVL rules up-to-date requires expert list curators to continuously monitor sites known to employ CV services and to discover new such sites in the wild-both tasks require considerable manual effort. To help automate and scale ACVL curation, we develop CV-INSPECTOR, a machine learning approach for automatically detecting adblock circumvention using differential execution analysis. We show that CV-INSPECTOR achieves 93% accuracy in detecting sites that successfully circumvent adblockers. We deploy CV-INSPECTOR on top-20K sites to discover the sites that employ circumvention in the wild. We further apply CV-INSPECTOR to a list of sites that are known to utilize circumvention and are closely monitored by ACVL authors. We demonstrate that CV-INSPECTOR reduces the human labeling effort by 98%, which removes a major bottleneck for ACVL authors. Our work is the first large-scale study of the state of the adblock circumvention arms race, and makes an important step towards automating anti-CV efforts.

AutoFR: Automated Filter Rule Generation for Adblocking

“…The research community has looked into the filter list curation process to investigate its effectiveness and painpoints [6,47,85,86]. Snyder et al [85] studied EasyList's evolution and showed that it needs to be frequently updated (median update interval of 1.12 hours) because of the dynamic nature of online advertising and efforts from advertisers to evade filter rules.…”

“…There is, however, optimism in using ML-based approaches to assist with maintenance of filter lists, in a off-line manner. For example, Brave [14], Adblock Plus [66], and the research community [47], have been using ML models to assist FL authors in prioritizing filter rule updates. However, they have two main limitations.…”

“…Filter List Authors' Workflow for Creating Filter Rules. Our design of AutoFR is motivated by the bottlenecks of filter rule generation, revealed by prior work [6,47], our discussions with FL authors, and our own experience in curating filter rules. Next, we break down the process that FL authors employ into a sequence of tasks, also illustrated in Fig.…”

“…There are hundreds of different adblocking filter lists that target different platforms and geographic regions [23]. It is well-known that the filter list curation process is slow and error-prone [6], and requires significant continuous effort by the filter list community to keep them up-to-date [47].…”

“…Fig. 1(a) shows the workflow of a FL author, which involves roughly the following steps: (1) select a network request to block; (2) design a filter rule that corresponds to this request and apply it on the site; (3) visually inspect the page to evaluate if the filter rule blocks ads and/or causes breakage and; (4) repeat for other network requests and rules; since modern sites are highly dynamic, and often more so in response to adblocking [6,18,47,91], the FL author may also need to revisit the site multiple times to ensure the rule is effective; and (5) stop when a set of filter rules can adequately block ads without causing breakage.…”