mlsanityData to 7 Oct 2026

Stepwise Reasoning Error Disruption Attack of LLMs

arXiv 2412.11934 ↗First posted 16 Dec 2024cs.AIcited by 8

On the map: LLM reasoning → Reasoning for agents, planning and code

Abstract (arXiv)

Large language models (LLMs) have made remarkable strides in complex reasoning tasks, but their safety and robustness in reasoning processes remain underexplored. Existing attacks on LLM reasoning are constrained by specific settings or lack of imperceptibility, limiting their feasibility and generalizability. To address these challenges, we propose the Stepwise rEasoning Error Disruption (SEED) attack, which subtly injects errors into prior reasoning steps to mislead the model into producing incorrect subsequent reasoning and final answers. Unlike previous methods, SEED is compatible with zero-shot and few-shot settings, maintains the natural reasoning flow, and ensures covert execution without modifying the instruction. Extensive experiments on four datasets across four different models demonstrate SEED's effectiveness, revealing the vulnerabilities of LLMs to disruptions in reasoning processes. These findings underscore the need for greater attention to the robustness of LLM reasoning to ensure safety in practical applications. Our code is available at: https://github.com/Applied-Machine-Learning-Lab/SEED-Attack.

Key facts. Cited by 8 papers in the map (data through 7 Oct 2026). Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning. Builds on 37 of its 53 references that are in the map. First posted on arXiv 16 Dec 2024.

Builds on

37 of 53 references.

Closest ancestors · 2

Same direction, newest first

  1. 2024Qwen2.5-Coder Technical Report18 Sep
  2. 2024BadChain: Backdoor Chain-of-Thought Prompting for Large Language Models20 Jan

Other references · 35

  1. 2023GPT-4 Technical Reportcited by 18,132
  2. 2024The Llama 3 Herd of Modelscited by 14,192
  3. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  4. 2022Self-Consistency Improves Chain of Thought Reasoning in Language Modelscited by 5,601
  5. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,203
  6. 2022Large Language Models are Zero-Shot Reasonerscited by 4,323
  7. 2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,317
  8. 2018CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledgecited by 1,793
  9. 2022Least-to-Most Prompting Enables Complex Reasoning in Large Language Modelscited by 1,397
  10. 2024A Survey on LLM-as-a-Judgecited by 1,384
  11. 2023Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Promptingcited by 1,081
  12. 2023Graph of Thoughts: Solving Elaborate Problems with Large Language Modelscited by 873
23 more
  1. 2022Automatic Chain of Thought Prompting in Large Language Modelscited by 684
  2. 2023Measuring Faithfulness in Chain-of-Thought Reasoningcited by 668
  3. 2019MathQA: Towards Interpretable Math Word Problem Solving with Operation-Based Formalismscited by 653
  4. 2024ReST-MCTS*: LLM Self-Training via Process Reward Guided Tree Searchcited by 338
  5. 2024Jailbreak Attacks and Defenses Against Large Language Models: A Surveycited by 215
  6. 2022On Second Thought, Let's Not Think Step by Step! Bias and Toxicity in Zero-Shot Reasoningcited by 191
  7. 2023Large Language Models for Generative Information Extraction: A Surveycited by 183
  8. 2024Marco-o1: Towards Open Reasoning Models for Open-Ended Solutionscited by 142
  9. 2023Automatic Prompt Augmentation and Selection with Chain-of-Thought from Labeled Datacited by 119
  10. 2023An LLM can Fool Itself: A Prompt-Based Adversarial Attackcited by 111
  11. 2023Skeleton-of-Thought: Prompting LLMs for Efficient Parallel Generationcited by 103
  12. 2024Q*: Improving Multi-step Reasoning for LLMs with Deliberative Planningcited by 95
  13. 2023Backdoor Attacks for In-Context Learning with Language Modelscited by 92
  14. 2024Buffer of Thoughts: Thought-Augmented Reasoning with Large Language Modelscited by 91
  15. 2023Prompt as Triggers for Backdoor Attack: Examining the Vulnerability in Language Modelscited by 87
  16. 2024Improved Few-Shot Jailbreaking Can Circumvent Aligned Language Models and Their Defensescited by 68
  17. 2024A Comparative Study on Reasoning Patterns of OpenAI's o1 Modelcited by 62
  18. 2022Exploring the Universal Vulnerability of Prompt-based Learning Paradigmcited by 56
  19. 2022Knowledge-enhanced Black-box Attacks for Recommendationscited by 25
  20. 2023Agent4Ranking: Semantic Robust Ranking via Personalized Query Rewriting Using Multi-agent LLMcited by 18
  21. 2024Preemptive Answer "Attacks" on Chain-of-Thought Reasoningcited by 16
  22. 2024Towards Personalized Evaluation of Large Language Models with An Anonymous Crowd-Sourcing Platformcited by 10
  23. 2022Graph Adaptive Semantic Transfer for Cross-domain Sentiment Classificationcited by 6

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 4; 2026: 4.

8 papers.

  1. 2025Optimas: Optimizing Compound AI Systems with Globally Aligned Local Rewardscited by 23
  2. 2025Don't Take the Premise for Granted: Evaluating the Premise Critique Ability of Large Language Modelscited by 8
  3. 2025BadThink: Triggered Overthinking Attacks on Chain-of-Thought Reasoning in Large Language Modelscited by 7
  4. 2026TextResNet: Decoupling and Routing Optimization Signals in Compound AI Systems via Deep Residual Tuningcited by 3
  5. 2025ZeroED: Hybrid Zero-shot Error Detection through Large Language Model Reasoningcited by 2
  6. 2026"I May Not Have Articulated Myself Clearly": Diagnosing Dynamic Instability in LLM Reasoning at Inference Timecited by 1
  7. 2026Learning When to Sample: Confidence-Aware Selective Sampling for Efficient Chain-of-Thought Reasoningcited by 1
  8. 2026Implicit Reasoning Steering via Concept Chainingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Token-Efficient Prompt Injection Attack: Provoking Cessation in LLM Reasoning via Adaptive Token Compression
  2. 2025Process or Result? Manipulated Ending Tokens Can Mislead Reasoning LLMs to Ignore the Correct Reasoning Steps
  3. 2025Rethinking Reasoning: A Survey on Reasoning-based Backdoors in LLMs
  4. 2025Exploring LLM Reasoning Through Controlled Prompt Variations
  5. 2026Beyond Content Safety: Real-Time Monitoring for Reasoning Vulnerabilities in Large Language Models
  6. 2025POT: Inducing Overthinking in LLMs via Black-Box Iterative Optimization
  7. 2024To Err is Machine: Vulnerability Detection Challenges LLM Reasoning
  8. 2025Reasoning Introduces New Poisoning Attacks Yet Makes Them More Complicated