The paper investigates the vulnerability of the PGM-index to poisoning attacks, introducing PGM-attack—a method that sequentially inserts adversarial keys to inflate the number of segments in the index’s piecewise linear approximation. Experiments show that poisoning just 10% of keys can increase segment count—and thus index size—by up to 120×, with the attack achieving at least 52% of the theoretical optimum. The attack also transfers to other learned indexes, especially those based on PLA, highlighting an intrinsic weakness in the PGM-index’s optimization objective.
arXiv:2606. 05609v1 Announce Type: cross Abstract: As large language models (LLMs) are widely deployed, identifying their vulnerability through jailbreak attacks becomes increasingly critical.
arXiv:2606. 08372v1 Announce Type: cross Abstract: Synthetic data is increasingly promoted as a privacy-preserving substitute for releasing sensitive tabular records, yet its central adversarial threat ("reconstruction", the recovery of an individual's hidden attribute values from a synthetic release and a handful of known quasi-identifiers) has been studied only in scattered, hard-to-compare settings.
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.
arXiv:2606. 11409v1 Announce Type: cross Abstract: Adversarial robustness evaluations of large language models (LLMs) typically report attack success rate (ASR) under fixed query budgets, implicitly treating all attacks as equally costly.
arXiv:2610.17382v1 Announce Type: cross Abstract: Score-based black-box adversarial attacks pose a serious threat to deployed machine learning systems by exploiting repeated model queries to craft ad...
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.
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.
arXiv:2608. 06477v1 Announce Type: cross Abstract: Computer-use agents (CUAs) face a growing threat from indirect prompt injection, where adversarial instructions are planted in the environment such as web pages.
arXiv:2607. 19855v1 Announce Type: new Abstract: Adversarial robustness is commonly evaluated with predefined attack ensembles, such as AutoAttack, at a single perturbation budget $\varepsilon$ and on a selective choice of perturbation norms.
The paper introduces Combinatorial Lock, a training‑free, hardware‑friendly defense against multi‑trigger backdoor attacks in AI models. It transforms multi‑trigger detection into a combinatorial problem over binary activation signatures, using anomaly‑gated spectral scrubbing and hash‑compressed signatures verified by expert ensembles. Experiments on six popular architectures show the method reduces attack success rates to 0.1% while maintaining 97.8% clean accuracy, with modest latency overhead on ASIC and FPGA platforms.
SecPO introduces a principled adversarial training framework to defend large language models against prompt injection attacks. By approximating an ideal injection‑invariant reference model and employing a new SecPO loss, the authors improve security while preserving utility. Combined with an efficient adversarial training pipeline that synthesizes strong injected inputs, the resulting SecPO + AT model significantly reduces attack success rates compared to existing defenses and maintains performance across multiple benchmarks.
arXiv:2503. 00065v4 Announce Type: replace-cross Abstract: Graph Neural Networks (GNNs) achieve high performance in various real-world applications, such as drug discovery, traffic states prediction, and recommendation systems.