Risk Under Pressure: Compute-Aware Evaluation of Adversarial Robustness in Language Models

• URL: http://arxiv.org/abs/2606.11409v1
• Date: Tue, 09 Jun 2026 19:59:12 GMT
• Title: Risk Under Pressure: Compute-Aware Evaluation of Adversarial Robustness in Language Models
• Abstract summary: Adrial robustness evaluations of large language models (LLMs) typically report attack success rate (ASR) under fixed query budgets.<n>We propose a compute-aware evaluation framework based on computational pressure, measured in cumulative floating-point operations.
• Score: 22.44538447178147
• License: http://creativecommons.org/licenses/by/4.0/
• 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. In practice, the computational expense of different attack strategies can vary by orders of magnitude. Consequently, ASR at a fixed budget can obscure the true effort required to jailbreak a model, thereby making it hard to determine whether an attack's cost justifies its payoff to the attacker. We propose a compute-aware evaluation framework based on computational pressure, measured in cumulative floating-point operations (FLOPs), as a proxy for adversarial effort. We introduce risk-compute curves, which map compute budgets to attack risk, and derive two metrics that summarize the average pressure required for a given attack to succeed. Across ten models spanning three families and four different stages in language model training and alignment, evaluated with three attack strategies (gradient-based, iterative refinement, and template-based) on two jailbreak robustness benchmarks, we find: (1) alignment training has non-monotonic effects on compute-space robustness; (2) scaling model size reduces gradient-based attack effectiveness but has limited impact on cheaper template-based attacks; (3) gradient-based attacks optimized on a surrogate model can transfer to a separate target model, providing a way to reduce attacker costs; (4) compute cost varies by up to ${\approx}5{\times}$ across harm categories within a single model; and (5) safety-aligned RL increases aggregate cost while leaving some categories disproportionately accessible. We release our framework to enable compute-aware risk assessment and evaluation.

Related papers

• $V_{0.5}$: Generalist Value Model as a Prior for Sparse RL Rollouts [81.48669089692189]
  Generalist Value Models (such as $V_0.5$) achieve pre-trained value estimation by explicitly encoding model capabilities in-context.<n>We propose $V_0.5$, which adaptively fuses the baseline predicted by such value model with the empirical mean derived from sparse rollouts.<n>$V_0.5$ significantly outperforms GRPO and DAPO, achieving faster convergence and over some 10% performance improvement.
  arXiv  Detail & Related papers  (2026-03-11T14:57:41Z)
• Adversarial Training for Defense Against Label Poisoning Attacks [53.893792844055106]
  Label poisoning attacks pose significant risks to machine learning models.<n>We propose a novel adversarial training defense strategy based on support vector machines (SVMs) to counter these threats.<n>Our approach accommodates various model architectures and employs a projected gradient descent algorithm with kernel SVMs for adversarial training.
  arXiv  Detail & Related papers  (2025-02-24T13:03:19Z)
• Fast Proxies for LLM Robustness Evaluation [48.53873823665833]
  We compare the ability of fast proxy metrics to predict the real-world robustness of an LLM against a simulated attacker ensemble.<n>This allows us to estimate a model's robustness to computationally expensive attacks without requiring runs of the attacks themselves.
  arXiv  Detail & Related papers  (2025-02-14T11:15:27Z)
• Model Tampering Attacks Enable More Rigorous Evaluations of LLM Capabilities [49.09703018511403]
  Evaluations of large language model (LLM) risks and capabilities are increasingly being incorporated into AI risk management and governance frameworks.<n>Currently, most risk evaluations are conducted by designing inputs that elicit harmful behaviors from the system.<n>We propose evaluating LLMs with model tampering attacks which allow for modifications to latent activations or weights.
  arXiv  Detail & Related papers  (2025-02-03T18:59:16Z)
• An Interpretable N-gram Perplexity Threat Model for Large Language Model Jailbreaks [87.64278063236847]
  In this work, we propose a unified threat model for the principled comparison of jailbreak attacks.<n>Our threat model checks if a given jailbreak is likely to occur in the distribution of text.<n>We adapt popular attacks to this threat model, and, for the first time, benchmark these attacks on equal footing with it.
  arXiv  Detail & Related papers  (2024-10-21T17:27:01Z)
• Advancing Generalized Transfer Attack with Initialization Derived Bilevel Optimization and Dynamic Sequence Truncation [49.480978190805125]
  Transfer attacks generate significant interest for black-box applications. Existing works essentially directly optimize the single-level objective w.r.t. surrogate model. We propose a bilevel optimization paradigm, which explicitly reforms the nested relationship between the Upper-Level (UL) pseudo-victim attacker and the Lower-Level (LL) surrogate attacker.
  arXiv  Detail & Related papers  (2024-06-04T07:45:27Z)
• Group-based Robustness: A General Framework for Customized Robustness in the Real World [16.376584375681812]
  We find that conventional metrics measuring targeted and untargeted robustness do not appropriately reflect a model's ability to withstand attacks from one set of source classes to another set of target classes. We propose a new metric, termed group-based robustness, that complements existing metrics and is better-suited for evaluating model performance in certain attack scenarios. We show that with comparable success rates, finding evasive samples using our new loss functions saves by a factor as large as the number of targeted classes.
  arXiv  Detail & Related papers  (2023-06-29T01:07:12Z)
• The Space of Adversarial Strategies [6.295859509997257]
  Adversarial examples, inputs designed to induce worst-case behavior in machine learning models, have been extensively studied over the past decade. We propose a systematic approach to characterize worst-case (i.e., optimal) adversaries.
  arXiv  Detail & Related papers  (2022-09-09T20:53:11Z)
• A Unified Evaluation of Textual Backdoor Learning: Frameworks and Benchmarks [72.7373468905418]
  We develop an open-source toolkit OpenBackdoor to foster the implementations and evaluations of textual backdoor learning. We also propose CUBE, a simple yet strong clustering-based defense baseline.
  arXiv  Detail & Related papers  (2022-06-17T02:29:23Z)
• Adversarial Poisoning Attacks and Defense for General Multi-Class Models Based On Synthetic Reduced Nearest Neighbors [14.968442560499753]
  State-of-the-art machine learning models are vulnerable to data poisoning attacks. This paper proposes a novel model-free label-flipping attack based on the multi-modality of the data. Second, a novel defense technique based on the Synthetic Reduced Nearest Neighbor (SRNN) model is proposed.
  arXiv  Detail & Related papers  (2021-02-11T06:55:40Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.