Shifting from Injection to Interaction: Rethinking Web Security in the Age of LLMs and Beyond

Abstract

Large language models (LLMs) are becoming integral to web applications and browser agents, transforming online interactions while introducing new attack vectors and reshaping longstanding web vulnerabilities. Classical threats such as cross-site scripting (XSS) can be amplified through LLM-mediated interactions, while LLM-specific vulnerabilities can propagate across web applications, introducing attacks such as prompt injection. Securing modern web systems therefore requires understanding inter

Categories

Framework mappings

Suggested from the entry's categories.

@misc{singh2026shifting,
  title = {{Shifting from Injection to Interaction: Rethinking Web Security in the Age of LLMs and Beyond}},
  author = {Nivedita Singh and Alsharif Abuadbba and Yansong Gao and Surya Nepal and Hyoungshick Kim},
  year = {2026},
  month = sep,
  eprint = {2609.03999},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.03999}
}