arXiv:2606. 03344v1 Announce Type: cross Abstract: Model merging composes specialized capabilities into a single LLM by aggregating task vectors sourced from unverified public platforms, exposing a critical supply-chain attack surface: Because any malicious behavior can be encoded into a task vector, and merging grants third-party vectors direct write access to model weights, an attacker-provided task vector can enable or amplify diverse downstream threats.
By Jinghuai Zhang, Yetian He, Kunlin Cai, Han Zhao, Fnu Suya, Yuan Tian
Model merging composes specialized capabilities into a single LLM by aggregating task vectors sourced from unverified public platforms, exposing a critical supply-chain attack surface: Because any malicious behavior can be encoded into a task vector, and merging grants third-party vectors direct write access to model weights, an attacker-provided task vector can enable or amplify diverse downstream threats. Prior work studies only backdoor attacks against model merging for classifiers using static arithmetic heuristics, which fail to effectively handle diverse attacks on generative LLMs for three reasons.
arXiv:2605. 15152v2 Announce Type: replace-cross Abstract: LLM quantization has become essential for memory-efficient deployment.
By Xiaohua Zhan, Kazuki Egashira, Robin Staab, Mark Vero, Martin Vechev
arXiv:2606. 28962v1 Announce Type: cross Abstract: Model quantization is essential for the efficient deployment of Large Language Models (LLMs), but introduces a critical vulnerability: Quantization-Conditioned Backdoor (QCB) attacks.
By Aoying Zheng, Anqi Du, Zizhuang Deng, Yuxuan Chen
arXiv:2606. 30360v1 Announce Type: new Abstract: The training-free integration of expert models via model merging has exposed significant security risks, enabling free-riders to combine specialized models without authorization.
By Kuangpu Guo, Qingyan Zheng, Jian Liang, Yongcan Yu, Zilei Wang, Ran He, Tieniu Tan
The paper introduces a new framework for creating frequency‑based adversarial attacks that are grounded in an explicit optimization problem. By defining a perturbation constraint set linked to structured, non‑orthogonal transforms, the authors show that attacks can be generated as weighted σ‒projections onto this set, providing a clear geometric characterization. Experiments on standard datasets demonstrate that these attacks are highly effective across both pretrained and robust models, even on unseen architectures.
By Vicky Kouni, Stelios Perrakis, Francis Bach, Pascal Frossard, Yann Chevaleyre
The paper introduces DYNASHIELD, a black-box moving target defense for large language models that dynamically customizes decoding hyperparameters and system prompts during inference. It first identifies decoding configurations that lower the probability of successful jailbreak attacks, then samples from a weighted pool of these configurations to add controlled variability to the model’s behavior. Experiments on seven open-source LLMs and four state-of-the-art jailbreak attacks show that DYNASHIELD significantly reduces attack success rates while preserving response quality and adding minimal inference overhead.
By Xiaoqun Liu, Weiming Qi, Qiben Yan
The paper examines how fine‑tuned defenses against prompt injection in large language models can be audited before deployment. By leveraging intermediate checkpoints from the fine‑tuning process, the authors create strong optimization‑based attacks (e.g., GCG and BEAST) that dramatically increase attack success rates—from near zero to up to 96%—against state‑of‑the‑art defenses. This demonstrates that current empirical evaluations often overestimate robustness and that pre‑deployment stress‑testing can uncover real vulnerabilities.
By Xiaoxue Yang, Bozhidar Stevanoski, Matthieu Meeus, Yves-Alexandre de Montjoye
The paper investigates how inference optimization for large language models can introduce numerical inconsistencies that trigger hidden backdoors. It introduces two types of optimization‑triggered backdoors: the Input‑Specific Optimization Backdoor (ISOB) and the Universal Optimization Backdoor (UOB), the latter enabling a model to remain benign under normal execution but activate a backdoor when optimization is applied. Experiments on seven open‑source LLMs, across multiple tasks and optimization backends, show UOB can achieve up to 100% attack success while maintaining clean accuracy, and the authors propose three defenses that reduce the attack success rate to 0.02.
By Yifei Wang, Yida Yang, Tianlin Li, Xiaohan Zhang, Xiaoyu Zhang, Li Pan
arXiv:2602. 06911v2 Announce Type: replace-cross Abstract: As increasingly capable open-weight large language models (LLMs) are deployed, improving their tamper resistance against unsafe modifications, whether accidental or intentional, becomes critical to minimize risks.
By Saad Hossain, Tom Tseng, Punya Syon Pandey, Samanvay Vajpayee, Matthew Kowal, Nayeema Nonta, Samuel Simko, Stephen Casper, Zhijing Jin, Kellin Pelrine, Sirisha Rambhatla
The paper introduces Quarantined Expert Shutdown (QES), a new backdoor containment strategy for large language models. QES allows backdoor learning to occur during training but routes it into a designated, quarantined expert that can be disabled at deployment. The method achieves significant reductions in attack success rates while largely preserving model utility.
By Jianwei Li, Min-Seon Kim, Jung-Eun Kim
arXiv:2609.14060v1 Announce Type: cross
Abstract: Quantization is one of the default deployment paths for open-weight LLM agents, but it is not behavior-preserving: an adversary can release a full-pr...
By Xiaoqun Liu, Qiben Yan