English
Related papers

Related papers: Colluding LoRA: A Compositional Vulnerability in L…

200 papers

Agents controlled by the output of reinforcement learning (RL) algorithms often transition to unsafe states, particularly in uncertain and partially observable environments. Partially observable Markov decision processes (POMDPs) provide a…

Systems and Control · Electrical Eng. & Systems 2025-09-16 Steven Carr , Georgios Bakirtzis , Ufuk Topcu

We propose a new formal criterion for evaluating secure compilation schemes for unsafe languages, expressing end-to-end security guarantees for software components that may become compromised after encountering undefined behavior---for…

Activation steering is a promising technique for controlling LLM behavior by adding semantically meaningful vectors directly into a model's hidden states during inference. It is often framed as a precise, interpretable, and potentially…

Machine Learning · Computer Science 2026-02-17 Anton Korznikov , Andrey Galichin , Alexey Dontsov , Oleg Y. Rogov , Ivan Oseledets , Elena Tutubalina

We show that LoRA adapters, the dominant distribution format for fine-tuned LLMs, can be reliably backdoored through training data poisoning while preserving baseline task performance. On a Qwen 2.5 1.5B prompt-injection classifier, a small…

Cryptography and Security · Computer Science 2026-05-29 Travis Lelle

Large language models (LLMs) have achieved widespread adoption across numerous applications. However, many LLMs are vulnerable to malicious attacks even after safety alignment. These attacks typically bypass LLMs' safety guardrails by…

Cryptography and Security · Computer Science 2025-06-17 Yucheng Li , Surin Ahn , Huiqiang Jiang , Amir H. Abdi , Yuqing Yang , Lili Qiu

Large language models (LLMs) have democratized software development, reducing the expertise barrier for programming complex applications. This accessibility extends to malicious software development, raising significant security concerns.…

Cryptography and Security · Computer Science 2025-07-04 Lu Yan , Zhuo Zhang , Xiangzhe Xu , Shengwei An , Guangyu Shen , Zhou Xuan , Xuan Chen , Xiangyu Zhang

Large-scale verifiable prompts underpin the success of Reinforcement Learning with Verifiable Rewards (RLVR), but they contain many uninformative examples and are costly to expand further. Recent studies focus on better exploiting limited…

Computation and Language · Computer Science 2026-04-23 Xin Xu , Clive Bai , Kai Yang , Tianhao Chen , Yangkun Chen , Weijie Liu , Hao Chen , Yang Wang , Saiyong Yang , Can Yang

In this study, we disclose a worrying new vulnerability in Large Language Models (LLMs), which we term \textbf{involuntary jailbreak}. Unlike existing jailbreak attacks, this weakness is distinct in that it does not involve a specific…

Cryptography and Security · Computer Science 2025-12-30 Yangyang Guo , Yangyan Li , Mohan Kankanhalli

Low rank adaptation (LoRA) has emerged as a prominent technique for fine-tuning large language models (LLMs) thanks to its superb efficiency gains over previous methods. While extensive studies have examined the performance and structural…

Machine Learning · Computer Science 2025-05-20 Zi Liang , Haibo Hu , Qingqing Ye , Yaxin Xiao , Ronghua Li

The increasing deployment of Large Language Models (LLMs) across enterprise and mission-critical domains has underscored the urgent need for robust guardrailing systems that ensure safety, reliability, and compliance. Existing solutions…

Computation and Language · Computer Science 2025-10-16 Karthik Avinash , Nikhil Pareek , Rishav Hada

Recent advancements in visual context compression enable MLLMs to process ultra-long contexts efficiently by rendering text into images. However, we identify a critical vulnerability inherent to this paradigm: lowering image resolution…

Computer Vision and Pattern Recognition · Computer Science 2026-05-11 Zhixue Song , Boyan Han , Yiwei Wang , Chi Zhang

Adversarial attacks can deceive neural networks by adding tiny perturbations to their input data. Ensemble defenses, which are trained to minimize attack transferability among sub-models, offer a promising research direction to improve…

Machine Learning · Computer Science 2022-11-16 Yunrui Yu , Xitong Gao , Cheng-Zhong Xu

Safety alignment in large language models (LLMs) is primarily evaluated under open-ended generation, where models can mitigate risk by refusing to respond. In contrast, many real-world applications place LLMs in structured decision-making…

Computation and Language · Computer Science 2026-04-21 Yuheng Chen , Zhiyu Wu , Bowen Cheng , Tetsuro Takahashi

While large language models (LLMs) such as Llama-2 or GPT-4 have shown impressive zero-shot performance, fine-tuning is still necessary to enhance their performance for customized datasets, domain-specific tasks, or other private needs.…

Machine Learning · Computer Science 2025-01-07 Chia-Yi Hsu , Yu-Lin Tsai , Chih-Hsun Lin , Pin-Yu Chen , Chia-Mu Yu , Chun-Ying Huang

Personality imbuing customizes LLM behavior, but safety evaluations almost always study prompt-based personas alone. We show this is incomplete: prompting and activation steering expose *different*, architecture-dependent vulnerability…

Artificial Intelligence · Computer Science 2026-04-15 Wenkai Li , Fan Yang , Shaunak A. Mehta , Koichi Onoue

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

Large language models (LLMs) have demonstrated superior performance compared to previous methods on various tasks, and often serve as the foundation models for many researches and services. However, the untrustworthy third-party LLMs may…

Cryptography and Security · Computer Science 2024-04-02 Hai Huang , Zhengyu Zhao , Michael Backes , Yun Shen , Yang Zhang

This study systematically analyzes the vulnerability of 36 large language models (LLMs) to various prompt injection attacks, a technique that leverages carefully crafted prompts to elicit malicious LLM behavior. Across 144 prompt injection…

Safety alignment mechanisms in Large Language Models (LLMs) often operate as latent internal states, obscuring the model's inherent capabilities. Building on this observation, we model the safety mechanism as an unobserved confounder from a…

Computation and Language · Computer Science 2026-02-09 Yao Zhou , Zeen Song , Wenwen Qiang , Fengge Wu , Shuyi Zhou , Changwen Zheng , Hui Xiong

We propose Orthogonal Monte Carlo Dropout, a mechanism that enforces strict orthogonality when combining sparse semantic vectors without extra time complexity. Low-Rank Adaptation (LoRA), a popular fine-tuning method for large models,…

Machine Learning · Computer Science 2025-10-09 Andi Zhang , Xuan Ding , Haofan Wang , Steven McDonagh , Samuel Kaski