Stealthy Backdoor Attacks against LLMs Based on Natural Style Triggers

Abstract

The growing application of large language models (LLMs) in safety-critical domains has raised urgent concerns about their security. Many recent studies have demonstrated the feasibility of backdoor attacks against LLMs. However, existing methods suffer from three key shortcomings: explicit trigger patterns that compromise naturalness, unreliable injection of attacker-specified payloads in long-form generation, and incompletely specified threat models that obscure how backdoors are delivered and

Categories

Framework mappings

Suggested from the entry's categories.

@misc{wei2026stealthy,
  title = {{Stealthy Backdoor Attacks against LLMs Based on Natural Style Triggers}},
  author = {Jiali Wei and Ming Fan and Guoheng Sun and Xicheng Zhang and Haijun Wang and Ting Liu},
  year = {2026},
  month = apr,
  eprint = {2604.21700},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2604.21700}
}