Researchers have introduced WebSwarm, a novel approach to recursive multi-agent orchestration, designed to enhance deep-and-wide web search capabilities. This method addresses the limitations of single-agent systems, which struggle to balance search depth and coverage due to constraints on trajectory length and context. By leveraging large language models (LLMs), WebSwarm enables multiple agents to collaborate and expand search coverage, making it particularly suited for complex research-oriented tasks. The development of WebSwarm has significant implications for information seeking, as it can potentially transform the way users interact with web search agents1. As LLMs continue to advance, their applications in web search will likely reshape the capability and risk surfaces of various systems, with security implications emerging in the wake of these developments. The introduction of WebSwarm highlights the need for practitioners to reevaluate their approaches to web search and consider the potential security consequences of relying on LLM-based agents.