EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

One-Prompt Censorship Evasion via Generative Diffusion Models

Shiyi Ling, Yuhang Gan, Chen Qian

Proceedings of the ACM on Networking · 19.11.2026 · 4 (CoNEXT4) · 1-20

DOI: 10.1145/3856970 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the ACM on Networking, online erschienen am 19.11.2026. Verfasst von Shiyi Ling und 2 weiteren Autor:innen. Die Arbeit nennt 47 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Science Foundation.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The escalating arms race between Internet censorship and evasion has driven censors to evolve from static rule-based filtering to sophisticated deep learning-based traffic analysis. While recent automated evasion tools have attempted to counter this by leveraging stochastic search and programmable heuristics, they continue to suffer from insufficient evasion robustness across diverse censorship modalities and poor usability due to complex, mechanism-specific configurations that require manual fitness tuning or domain-specific languages. In this paper, we propose a paradigm shift that reframes censorship evasion as a semantic image-to-image editing task, allowing users to execute it with a single prompt. We introduce FlowPaint, a novel generative framework that adapts large pretrained diffusion models to automatically reshape censored traffic into benign patterns. FlowPaint utilizes an instruction-tuned diffusion architecture to perform semantic editing on network flows. Evaluations against both industrial-grade rule-based middleboxes and learning-based classifiers demonstrate that FlowPaint outperforms existing censorship evasion baselines, enabling users to counter diverse censorship paradigms solely by varying natural language instructions.

Förderung

  • National Science Foundation

Zitieren

APA

Ling, S., Gan, Y., & Qian, C. (2026). One-Prompt Censorship Evasion via Generative Diffusion Models. Proceedings of the ACM on Networking, 4(CoNEXT4), 1-20. https://doi.org/10.1145/3856970

Vancouver

Ling S, Gan Y, Qian C. One-Prompt Censorship Evasion via Generative Diffusion Models. Proceedings of the ACM on Networking. 2026;4(CoNEXT4):1-20. doi:10.1145/3856970

BibTeX
@article{Ling2026,
  author = {Ling, Shiyi and Gan, Yuhang and Qian, Chen},
  title = {One-Prompt Censorship Evasion via Generative Diffusion Models},
  journal = {Proceedings of the ACM on Networking},
  year = {2026},
  volume = {4},
  number = {CoNEXT4},
  pages = {1--20},
  doi = {10.1145/3856970}
}
Kostenlos starten →