English
Related papers

Related papers: Thinking Wrong in Silence: Backdoor Attacks on Con…

200 papers

Backdoor attack poses a significant security threat to Deep Learning applications. Existing attacks are often not evasive to established backdoor detection techniques. This susceptibility primarily stems from the fact that these attacks…

Computer Vision and Pattern Recognition · Computer Science 2024-03-27 Siyuan Cheng , Guanhong Tao , Yingqi Liu , Guangyu Shen , Shengwei An , Shiwei Feng , Xiangzhe Xu , Kaiyuan Zhang , Shiqing Ma , Xiangyu Zhang

Large Language Models (LLMs) demonstrate their reasoning ability through chain-of-thought (CoT) generation. However, LLM's autoregressive decoding may limit the ability to revisit and refine earlier tokens in a holistic manner, which can…

Machine Learning · Computer Science 2026-04-24 Haoqiang Kang , Yizhe Zhang , Nikki Lijing Kuang , Nicklas Majamaki , Navdeep Jaitly , Yi-An Ma , Lianhui Qin

Binary vulnerability analysis is increasingly performed by LLM-based agents in an iterative, multi-pass manner, with the model as the core decision-maker. However, how such systems organize exploration over hundreds of reasoning steps…

Artificial Intelligence · Computer Science 2026-03-20 Qiang Li , XiangRui Zhang , Haining Wang

Current frontier large-language models rely on reasoning to achieve state-of-the-art performance. Many existing interpretability are limited in this area, as standard methods have been designed to study single forward passes of a model…

Machine Learning · Computer Science 2025-10-28 Paul C. Bogdan , Uzay Macar , Neel Nanda , Arthur Conmy

Recent reasoning large language models (LLMs) have demonstrated remarkable improvements in mathematical reasoning capabilities through long Chain-of-Thought. The reasoning tokens of these models enable self-correction within reasoning…

Artificial Intelligence · Computer Science 2025-04-02 Yu Cui , Bryan Hooi , Yujun Cai , Yiwei Wang

Large Reasoning Models (LRMs) improve task performance through extended inference-time reasoning. Although previous studies suggest that longer reasoning should lead to more robust safety behavior, we find evidence to the contrary:…

Artificial Intelligence · Computer Science 2026-05-26 Jianli Zhao , Tingchen Fu , Rylan Schaeffer , Mrinank Sharma , Fazl Barez

Mitigating explicit and implicit biases in Large Language Models (LLMs) has become a critical focus in the field of natural language processing. However, many current methodologies evaluate scenarios in isolation, without considering the…

Computation and Language · Computer Science 2025-02-04 Robert Morabito , Sangmitra Madhusudan , Tyler McDonald , Ali Emami

Chain-of-thought (CoT) reasoning has been proposed as a transparency mechanism for large language models in safety-critical deployments, yet its effectiveness depends on faithfulness (whether models accurately verbalize the factors that…

Computation and Language · Computer Science 2026-03-25 Richard J. Young

Consistency models are a new class of models that generate images by directly mapping noise to data, allowing for one-step generation and significantly accelerating the sampling process. However, their robustness against adversarial attacks…

Cryptography and Security · Computer Science 2025-02-18 Chengen Wang , Murat Kantarcioglu

We introduce the State Stream Transformer (SST), a novel LLM architecture that reveals emergent reasoning behaviours and capabilities latent in pretrained weights through addressing a fundamental limitation in traditional transformer…

Machine Learning · Computer Science 2025-01-31 Thea Aviss

Developers increasingly construct multimodal large language models (MLLMs) by assembling pretrained components,introducing supply-chain attack surfaces.Existing security research primarily focuses on poisoning backbones such as encoders or…

Cryptography and Security · Computer Science 2026-05-11 Runhe Wang , Li Bai , Haibo Hu , Songze Li

Supervised fine-tuning has become the predominant method for adapting large pretrained models to downstream tasks. However, recent studies have revealed that these models are vulnerable to backdoor attacks, where even a small number of…

Cryptography and Security · Computer Science 2025-01-08 Peihai Jiang , Xixiang Lyu , Yige Li , Jing Ma

Backdoor (trojan) attacks embed hidden, controllable behaviors into machine-learning models so that models behave normally on benign inputs but produce attacker-chosen outputs when a trigger is present. This survey reviews the rapidly…

Cryptography and Security · Computer Science 2025-09-10 Bilal Hussain Abbasi , Yanjun Zhang , Leo Zhang , Shang Gao

Large Language Model interfaces are increasingly verbose, exposing intermediate reasoning traces alongside final answers. Traces are framed as transparency mechanisms, yet it is unclear how people use them to solve problems. We report a…

Human-Computer Interaction · Computer Science 2026-05-26 Daniela Fernandes , Daniel Buschek , Lev Tankelevitch , Thomas Kosch , Robin Welsch

Should LLM reasoning live in a separate module, or within a single model's forward pass and representational space? We study dual-architecture latent reasoning, where a fluent Base exchanges latent messages with a Coprocessor, and test two…

Fine-tuned Large Language Models (LLMs) are vulnerable to backdoor attacks through data poisoning, yet the internal mechanisms governing these attacks remain a black box. Previous research on interpretability for LLM safety tends to focus…

Cryptography and Security · Computer Science 2025-10-01 Miao Yu , Zhenhong Zhou , Moayad Aloqaily , Kun Wang , Biwei Huang , Stephen Wang , Yueming Jin , Qingsong Wen

Discrete diffusion language models (dLLMs) generate text by iteratively denoising a masked sequence. However, standard dLLMs condition each denoising step solely on the current hard-masked sequence, while intermediate continuous…

Computation and Language · Computer Science 2026-03-31 Kejing Xia , Mingzhe Li , Lixuan Wei , Zhenbang Du , Xiangchi Yuan , Dachuan Shi , Qirui Jin , Wenke Lee

Extended-thinking models expose a second text-generation channel ("thinking tokens") alongside the user-visible answer. This study examines 12 open-weight reasoning models on MMLU and GPQA questions paired with misleading hints. Among the…

Computation and Language · Computer Science 2026-03-30 Richard J. Young

The rapid proliferation of open-source language models significantly increases the risks of downstream backdoor attacks. These backdoors can introduce dangerous behaviours during model deployment and can evade detection by conventional…

Modern language models remain vulnerable to backdoor attacks via poisoned data, where training inputs containing a trigger are paired with a target output, causing the model to reproduce that behavior whenever the trigger appears at…

Cryptography and Security · Computer Science 2026-01-06 Eric Xue , Ruiyi Zhang , Pengtao Xie