Metraipolis › archive › dc_34cc4fc13f0e
Diffusion Language Models for Perplexity-Aware Black-Box Attacks — arXiv:2610.31558
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.31558
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-25T08:36:48.315Z
- SHA-256 of the archived text
851f25506023a80f4c5e6166bca01265220da2eb5434100476821f562045b3a3- Stable URL of this page
https://metraipolis.com/documento/dc_34cc4fc13f0e
How to cite
Diffusion Language Models for Perplexity-Aware Black-Box Attacks — arXiv:2610.31558 — arXiv (en), https://arxiv.org/abs/2610.31558 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_34cc4fc13f0e (Metraipolis archive dc_34cc4fc13f0e, sha256 851f25506023a80f4c5e6166bca01265220da2eb5434100476821f562045b3a3)
BibTeX
@misc{dc34cc4fc13f0e,
title = {Diffusion Language Models for Perplexity-Aware Black-Box Attacks — arXiv:2610.31558},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.31558},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_34cc4fc13f0e; sha256 851f25506023a80f4c5e6166bca01265220da2eb5434100476821f562045b3a3},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.31558. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
\emph{Continuous-space} attacks find prompts that trigger malicious behaviors, such as overthinking, in black-box LLMs by sampling or optimizing over a low-dimensional Gaussian latent space and decoding it to text through a \emph{random mapping} and nearest-token lookup. Because the random mapping is not learned from language, its prompts can be highly unnatural, and as such easily detectable through input guardrails, and the risk estimates they yield are not grounded in natural prompts. In this paper, we demonstrate that the same continuous latent-space framework can be coupled directly to a learned generative distribution over language, a continuous diffusion language model (dLLM), which removes the artificial random mapping while preserving the standard-Gaussian latent formulation and the downstream sampling or optimization procedure. We evaluate this approach under two settings: sampling-based estimation using \emph{RareTrap} and optimization-based search using \emph{ThinkTrap}. We find that character-level perplexity drops from $10^3$--$10^7$ to 3--5 in RareTrap and from 20--28 to 4.3--5.4 in ThinkTrap. We also find that model risk estimates can differ by an order of magnitude across the two mappings.
Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Computation and Language (cs.CL); Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2610.31558 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-25T08:31:27.847Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).