Reproduce “Dynamics of Adversarial Attacks on Large Language Model-Based Search Engines” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xiyang Hu
The authors model LLM search competition as an infinitely repeated game where two providers choose whether to manipulate rankings, with attack success, cost, market degradation, and patience shaping payoffs. Under Grim Trigger, cooperation can persist above a threshold. In this theoretical model, lower success can locally raise defection incentives, while a cap may leave the best payoff unchanged under specified conditions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.