Metraipolis › archive › dc_7f5a2b455bec

Assessing Adversarial Robustness of Latent Reasoning Models — arXiv:2609.22228

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.22228
License (as archived)
CC0-1.0
Archived
2026-10-09T06:49:22.392Z
SHA-256 of the archived text
bd115ce4e8995e518e1035d869e76b17f179e42f3132d4f6730fcb321d8c8074
Stable URL of this page
https://metraipolis.com/documento/dc_7f5a2b455bec

How to cite

Assessing Adversarial Robustness of Latent Reasoning Models — arXiv:2609.22228 — arXiv (en), https://arxiv.org/abs/2609.22228 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7f5a2b455bec (Metraipolis archive dc_7f5a2b455bec, sha256 bd115ce4e8995e518e1035d869e76b17f179e42f3132d4f6730fcb321d8c8074)
BibTeX
@misc{dc7f5a2b455bec,
  title = {Assessing Adversarial Robustness of Latent Reasoning Models — arXiv:2609.22228},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.22228},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7f5a2b455bec; sha256 bd115ce4e8995e518e1035d869e76b17f179e42f3132d4f6730fcb321d8c8074},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.22228. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models increasingly rely on long chain-of-thought (CoT) trajectories for complex reasoning, but autoregressive generation brings substantial memory and inference costs. Latent reasoning models (LRMs) offer a more efficient alternative by compressing intermediate reasoning into a small number of continuous latent vectors. Despite their efficiency, however, the adversarial robustness of LRMs remains largely underexplored. In this work, we systematically evaluate the robustness of latent reasoning across textual and multimodal settings, covering eight models and six benchmarks. We find that, across our evaluated settings, LRMs are generally less robust than explicit CoT baselines under adversarial perturbations, with particularly severe degradation under white-box attacks. Further analysis reveals distinct failure modes across modalities: textual latent states exhibit brittle dynamics and high sensitivity to specific input patterns, while latent states in multimodal models can remain largely invariant to input perturbations and have limited influence on final predictions. These findings expose robustness limitations of current latent reasoning approaches and highlight the need to jointly consider efficiency and robustness when designing implicit reasoning systems. We have open-sourced our code to facilitate reproduction of our research this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-22 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG); Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2609.22228 — 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 2026-10-09T06:48:44.904Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-22?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).