mlsanityData to 14 Oct 2026

On the Efficiency-Safety Dilemma in Large Reasoning Models

arXiv 2609.23587 ↗First posted 20 Sep 2026cs.CL cs.AIcited by 0

On the map: LLM reasoning → Reasoning for agents, planning and code also fits: Test-time compute and efficient thinking

also in 1 more
Abstract (arXiv)

Large reasoning models (LRMs) incur high inference costs, often mitigated by efficiency techniques like quantization and pruning. However, the impact of these techniques on model adversarial robustness remains largely unexplored. This study provides the first comprehensive analysis of the interplay between efficiency, jailbreak vulnerability, and reasoning in LRMs. We find that while efficiency methods seemingly reduce the success rate of jailbreak attacks, this improvement is often superficial. It largely arises from degraded reasoning capabilities leading to "attempted but failed" malicious responses, rather than an increase in genuine alignment. Mechanistic analysis of representational drift confirms this, revealing a strict coupling between reasoning capability loss and the model's inability to maintain malicious semantic trajectories. Additionally, we identify quantization with pruning as the optimal strategy to balance efficiency and robustness. These findings clarify the distinction between true safety alignment and capability-induced failure, providing an empirical foundation for LRM deployment.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning. Builds on 32 of its 58 references that are in the map. First posted on arXiv 20 Sep 2026.

Builds on

32 of 58 references.

Closest ancestors · 8

Same direction, newest first

  1. 2026Reasoning Structure Matters for Safety Alignment of Reasoning Models21 Apr
  2. 2026Towards Safer Large Reasoning Models by Promoting Safety Decision-Making before Chain-of-Thought Generation18 Mar
  3. 2025Safety in Large Reasoning Models: A Survey24 Apr
  4. 2025Cats Confuse Reasoning LLM: Query Agnostic Adversarial Triggers for Reasoning Models3 Mar
  5. 2025A Mousetrap: Fooling Large Reasoning Models for Jailbreak with Chain of Iterative Chaos19 Feb
  6. 2025Investigating the Impact of Quantization Methods on the Safety and Reliability of Large Language Models18 Feb
  7. 2025H-CoT: Hijacking the Chain-of-Thought Safety Reasoning Mechanism to Jailbreak Large Reasoning Models, Including OpenAI o1/o3, DeepSeek-R1, and Gemini 2.0 Flash Thinking18 Feb
  8. 2025SafeChain: Safety of Language Models with Long Chain-of-Thought Reasoning Capabilities17 Feb

Other references · 24

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2024The Llama 3 Herd of Modelscited by 14,715
  3. 2025Qwen3 Technical Reportcited by 8,820
  4. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  5. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  6. 2023Mistral 7Bcited by 3,164
  7. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 3,104
  8. 2022GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformerscited by 1,919
  9. 2023A Simple and Effective Pruning Approach for Large Language Modelscited by 800
  10. 2023AutoDAN: Generating Stealthy Jailbreak Prompts on Aligned Large Language Modelscited by 785
  11. 2023Catastrophic Jailbreak of Open-source LLMs via Exploiting Generationcited by 451
  12. 2024PyramidKV: Dynamic KV Cache Compression based on Pyramidal Information Funnelingcited by 432
12 more
  1. 2023Sheared LLaMA: Accelerating Language Model Pre-training via Structured Pruningcited by 415
  2. 2023Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models' Alignmentcited by 401
  3. 2024SliceGPT: Compress Large Language Models by Deleting Rows and Columnscited by 385
  4. 2023Multi-step Jailbreaking Privacy Attacks on ChatGPTcited by 333
  5. 2023A Wolf in Sheep's Clothing: Generalized Nested Jailbreak Prompts can Fool Large Language Models Easilycited by 259
  6. 2023Removing RLHF Protections in GPT-4 via Fine-Tuningcited by 171
  7. 2024LaCo: Large Language Model Pruning via Layer Collapsecited by 126
  8. 2024No Token Left Behind: Reliable KV Cache Compression via Importance-Aware Mixed Precision Quantizationcited by 101
  9. 2024Decoding Compressed Trust: Scrutinizing the Trustworthiness of Efficient LLMs Under Compressioncited by 57
  10. 2024KVSharer: Efficient Inference via Layer-Wise Dissimilar KV Cache Sharingcited by 41
  11. 2024Fine-Tuning, Quantization, and LLMs: Navigating Unintended Outcomescited by 14
  12. 2024HarmLevelBench: Evaluating Harm-Level Compliance and the Impact of Quantization on Model Alignmentcited by 9

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024Reasoning as a Weapon: Adaptive Dual-Path Jailbreak Attack on Large Language Models
  2. 2025SafeMLRM: Demystifying Safety in Multi-modal Large Reasoning Models
  3. 2025HauntAttack: When Attack Follows Reasoning as a Shadow
  4. 2026Consistency of Large Reasoning Models Under Multi-Turn Attacks
  5. 2026Reasoning Hijacking: The Fragility of Reasoning Alignment in Large Language Models
  6. 2025Weakest Link in the Chain: Security Vulnerabilities in Advanced Reasoning Models
  7. 2026Dual-Adversarial Safety Alignment: Cultivating Intrinsic Threat Comprehension in LRMs
  8. 2025ARMOR: Aligning Secure and Safe Large Language Models via Meticulous Reasoning