Metraipolis › archive › dc_8a2b267b49ae

Does Deeper Reasoning Compromise Alignment? Revealing and Mitigating of Alignment Collapse in Large Reasoning Models — arXiv:2609.08186

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.08186
License (as archived)
CC0-1.0
Archived
2026-10-16T12:57:52.005Z
SHA-256 of the archived text
b2af13f45f780cffee1d879b05c38dda4b24e4e7c8b0911531fa9af3bfe6316c
Stable URL of this page
https://metraipolis.com/documento/dc_8a2b267b49ae

How to cite

Does Deeper Reasoning Compromise Alignment? Revealing and Mitigating of Alignment Collapse in Large Reasoning Models — arXiv:2609.08186 — arXiv (en), https://arxiv.org/abs/2609.08186 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_8a2b267b49ae (Metraipolis archive dc_8a2b267b49ae, sha256 b2af13f45f780cffee1d879b05c38dda4b24e4e7c8b0911531fa9af3bfe6316c)
BibTeX
@misc{dc8a2b267b49ae,
  title = {Does Deeper Reasoning Compromise Alignment? Revealing and Mitigating of Alignment Collapse in Large Reasoning Models — arXiv:2609.08186},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.08186},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_8a2b267b49ae; sha256 b2af13f45f780cffee1d879b05c38dda4b24e4e7c8b0911531fa9af3bfe6316c},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.08186. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

The emergence of Chain-of-Thought (CoT) has established a robust foundation for Large Reasoning Models (LRMs). While deep reasoning is widely believed to enhance safety alignment, the stability of alignment mechanisms under extended reasoning remains underexplored. This paper challenges the prevailing view by revealing a critical vulnerability: Deep Reasoning May Induce Alignment Collapse. To rigorously quantify this phenomenon, we propose the Alignment Loss Rate (ALR) metric. Our experiments demonstrate that as reasoning depth increases, ALR rises significantly, indicating a severe degradation in model robustness against external perturbations. Capitalizing on this instability, a novel jailbreaking paradigm, Reasoning Trap (RT), is proposed. RT induces the model into extended reasoning to amplify the impact of adversarial attacks, leading to a sharp decline in safety capabilities. To elucidate the mechanism behind this collapse, we identify Attention Dilution as the root cause, arising from the competition for attention between the extended reasoning process and the original input. To mitigate this, Reasoning Residual Alignment (RRA), a lightweight defense strategy that dynamically re-emphasizes the input via residual connections integrated with the reasoning process.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-09 in cs.CL).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.08186 — 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 2026-10-16T12:30:25.341Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).