Using AI Agents to Automate Black-Box Audits of Personalization Algorithms at Scale

• URL: http://arxiv.org/abs/2606.30801v1
• Date: Mon, 29 Jun 2026 18:25:09 GMT
• Title: Using AI Agents to Automate Black-Box Audits of Personalization Algorithms at Scale
• Abstract summary: Personalization algorithms determine what content users encounter on online platforms.<n> auditing these systems is difficult because independent auditors have only black-box access to the algorithms.<n>We introduce a framework for black-box audits of personalization algorithms using generative AI agents as behavioral engines for synthetic accounts.
• Score: 76.72846830885882
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Personalization algorithms determine what content users encounter on online platforms. Auditing these systems is difficult because independent auditors have only black-box access to the algorithms, while personalization depends on users' attributes, behavior, and evolving interaction histories. Existing auditing methods face a tradeoff: studies with real users capture realistic behavior but are costly and hard to control, whereas sock-puppet audits scale more easily but often rely on scripted behavior that limits realism. Beyond this, both approaches struggle to decouple user attributes from user behavior, limiting our ability to causally understand personalization. To address this gap, we introduce a framework for black-box audits of personalization algorithms using generative AI agents as behavioral engines for synthetic accounts. Each agent is instantiated with a fixed persona, grounded in demographic and political survey data, and interacts with a platform's content by reasoning about it and choosing actions. Because behavior is fixed within each persona while platform-visible signals such as age, gender, or location can be experimentally perturbed, our design enables counterfactual auditing of how platforms respond to user attributes. As a case study, we deploy 1,120 agents on X shortly after the 2024 U.S. election, spanning 14 personas and three counterfactual conditions, collecting over 200,000 content exposures. We find that X's algorithmic feed amplifies toxic, polarizing, political, and right-leaning content relative to the chronological feed, with amplification varying sharply by user ideology. Counterfactual analyses show that demographic signals affect content delivery in persona-dependent ways: pooled effects are largely null, while subgroup-level effects vary in direction and magnitude. Our work establishes GenAI-based agents as a new tool for algorithmic auditing.

Related papers

• How to Interpret Agent Behavior [56.59836196946289]
  We introduce ACT*ONOMY, a taxonomy for describing and analyzing agent behavior at runtime.<n>By providing a shared vocabulary, ACT*ONOMY helps researchers, agent designers, and end users interpret agent behavior more consistently.
  arXiv  Detail & Related papers  (2026-05-13T14:52:40Z)
• KnowU-Bench: Towards Interactive, Proactive, and Personalized Mobile Agent Evaluation [72.01173512175531]
  KnowU-Bench is an online benchmark for personalized mobile agents.<n>It covers 42 general GUI tasks, 86 personalized tasks, and 64 proactive tasks.<n>Our experiments reveal a striking degradation: agents that excel at explicit task execution fall below 50% under vague instructions.
  arXiv  Detail & Related papers  (2026-04-09T16:50:50Z)
• Beyond Content: Behavioral Policies Reveal Actors in Information Operations [1.1693100084511296]
  We introduce a platform-agnostic framework that identifies malicious actors from their behavioral policies by modeling user activity as sequential decision processes.<n>We apply this approach to 12,064 Reddit users, including 99 accounts linked to the Russian Internet Research Agency in Reddit's 2017 transparency report.<n>Activity-based representations, which model how users act rather than what they post, consistently outperform content models in detecting malicious accounts.
  arXiv  Detail & Related papers  (2026-02-02T21:39:21Z)
• Measuring Strategization in Recommendation: Users Adapt Their Behavior to Shape Future Content [66.71102704873185]
  We test for user strategization by conducting a lab experiment and survey. We find strong evidence of strategization across outcome metrics, including participants' dwell time and use of "likes" Our findings suggest that platforms cannot ignore the effect of their algorithms on user behavior.
  arXiv  Detail & Related papers  (2024-05-09T07:36:08Z)
• User Strategization and Trustworthy Algorithms [81.82279667028423]
  We show that user strategization can actually help platforms in the short term. We then show that it corrupts platforms' data and ultimately hurts their ability to make counterfactual decisions.
  arXiv  Detail & Related papers  (2023-12-29T16:09:42Z)
• Sociotechnical Audits: Broadening the Algorithm Auditing Lens to Investigate Targeted Advertising [1.6422305456306596]
  We propose the concept of sociotechnical auditing, focusing on the interplay between algorithms and users as each impacts the other. We develop Intervenr, a platform for conducting longitudinal sociotechnical audits with consenting, compensated participants. We deploy Intervenr in a two-week sociotechnical audit of online advertising to investigate the central premise that personalized ad targeting is more effective on users.
  arXiv  Detail & Related papers  (2023-08-30T05:26:47Z)
• Mathematical Framework for Online Social Media Auditing [5.384630221560811]
  Social media platforms (SMPs) leverage algorithmic filtering (AF) as a means of selecting the content that constitutes a user's feed with the aim of maximizing their rewards. Selectively choosing the contents to be shown on the user's feed may yield a certain extent of influence, either minor or major, on the user's decision-making. We mathematically formalize this framework and utilize it to construct a data-driven statistical auditing procedure to regulate AF from deflecting users' beliefs over time, along with sample complexity guarantees.
  arXiv  Detail & Related papers  (2022-09-12T19:04:14Z)
• Modeling Content Creator Incentives on Algorithm-Curated Platforms [76.53541575455978]
  We study how algorithmic choices affect the existence and character of (Nash) equilibria in exposure games. We propose tools for numerically finding equilibria in exposure games, and illustrate results of an audit on the MovieLens and LastFM datasets.
  arXiv  Detail & Related papers  (2022-06-27T08:16:59Z)
• Not all users are the same: Providing personalized explanations for sequential decision making problems [25.24098967133101]
  This work proposes an end-to-end adaptive explanation generation system. It begins by learning the different types of users that the agent could interact with. It is then tasked with identifying the type on the fly and adjust its explanations accordingly.
  arXiv  Detail & Related papers  (2021-06-23T07:46:19Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.