Does Deeper Reasoning Compromise Alignment? Revealing and Mitigating of Alignment Collapse in Large Reasoning Models

• URL: http://arxiv.org/abs/2609.08186v1
• Date: Tue, 08 Sep 2026 03:18:22 GMT
• Title: Does Deeper Reasoning Compromise Alignment? Revealing and Mitigating of Alignment Collapse in Large Reasoning Models
• Abstract summary: This paper challenges the prevailing view by revealing a critical vulnerability: Deep Reasoning May Induce Alignment Collapse.<n>Our experiments demonstrate that as reasoning depth increases, ALR rises significantly, indicating a severe degradation in model robustness against external perturbations.<n>We identify Attention Dilution as the root cause, arising from the competition for attention between the extended reasoning process and the original input.
• Score: 8.037716792115782
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: 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.

Related papers

• Alignment-Weighted DPO: A principled reasoning approach to improve safety alignment [13.463606100715504]
  Large language models are vulnerable to attacks that disguise harmful intent.<n>This vulnerability stems from shallow alignment mechanisms that lack deep reasoning.<n>We propose enhancing alignment through reasoning-aware post-training.
  arXiv  Detail & Related papers  (2026-02-24T20:30:51Z)
• Mitigating Safety Tax via Distribution-Grounded Refinement in Large Reasoning Models [63.368505631152594]
  Safety alignment incurs safety tax that perturbs a large reasoning model's (LRM) general reasoning ability.<n>Existing datasets used for safety alignment for an LRM are usually constructed by distilling safety reasoning traces and answers from an external LRM or human labeler.<n>We propose a safety alignment dataset construction method, dubbed DGR. DGR transforms and refines an existing out-of-distributional safety reasoning dataset to be aligned with the target's LLM inner distribution.
  arXiv  Detail & Related papers  (2026-02-02T14:18:48Z)
• APR: Penalizing Structural Redundancy in Large Reasoning Models via Anchor-based Process Rewards [61.52322047892064]
  Test-Time Scaling (TTS) has significantly enhanced the capabilities of Large Reasoning Models (LRMs)<n>We observe that LRMs frequently conduct repetitive self-verification without revision even after obtaining the final answer during the reasoning process.<n>We propose Anchor-based Process Reward (APR), a structure-aware reward shaping method that localizes the reasoning anchor and penalizes exclusively the post-anchor AST.
  arXiv  Detail & Related papers  (2026-01-31T14:53:20Z)
• THINKSAFE: Self-Generated Safety Alignment for Reasoning Models [60.10077024249373]
  We propose ThinkSafe, a framework that restores safety alignment without external teachers.<n>Our key insight is that while compliance suppresses safety mechanisms, models often retain latent knowledge to identify harm.<n> Experiments on DeepSeek-R1-Distill and Qwen3 show ThinkSafe significantly improves safety while preserving reasoning proficiency.
  arXiv  Detail & Related papers  (2026-01-30T16:31:02Z)
• Consistency Is Not Always Correct: Towards Understanding the Role of Exploration in Post-Training Reasoning [75.79451512757844]
  Foundation models exhibit broad knowledge but limited task-specific reasoning.<n> RLVR and inference scaling motivate post-training strategies such as RLVR and inference scaling.<n>We show that RLVR induces a squeezing effect, reducing reasoning entropy and forgetting some correct paths.
  arXiv  Detail & Related papers  (2025-11-10T18:25:26Z)
• AdvChain: Adversarial Chain-of-Thought Tuning for Robust Safety Alignment of Large Reasoning Models [62.70575022567081]
  We propose AdvChain, an alignment paradigm that teaches models dynamic self-correction through adversarial CoT tuning.<n>Our work establishes a new direction for building more robust and reliable reasoning models.
  arXiv  Detail & Related papers  (2025-09-29T04:27:23Z)
• When Thinking Backfires: Mechanistic Insights Into Reasoning-Induced Misalignment [23.096167213579957]
  Reasoning-Induced Misalignment (RIM) emerges when reasoning capabilities strengthen.<n>RIM is caused when specific types of reasoning patterns are introduced during inference or training.<n>During training, we find significantly higher activation entanglement between reasoning and safety in safety-critical neurons.
  arXiv  Detail & Related papers  (2025-08-30T16:04:54Z)
• Is Reasoning All You Need? Probing Bias in the Age of Reasoning Language Models [0.0]
  Reasoning Language Models (RLMs) have gained traction for their ability to perform complex, multi-step reasoning tasks.<n>While these capabilities promise improved reliability, their impact on robustness to social biases remains unclear.<n>We leverage the CLEAR-Bias benchmark to investigate the adversarial robustness of RLMs to bias elicitation.
  arXiv  Detail & Related papers  (2025-07-03T17:01:53Z)
• Lost at the Beginning of Reasoning [85.17612793300238]
  We show that the first reasoning step exerts a disproportionately large influence on the final prediction.<n>We propose an efficient sampling strategy that leverages a reward model to identify and retain high-quality first reasoning steps.
  arXiv  Detail & Related papers  (2025-06-27T09:53:57Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.