Capability-Based Scaling Laws for LLM Red-Teaming

• URL: http://arxiv.org/abs/2505.20162v1
• Date: Mon, 26 May 2025 16:05:41 GMT
• Title: Capability-Based Scaling Laws for LLM Red-Teaming
• Authors: Alexander Panfilov, Paul Kassianik, Maksym Andriushchenko, Jonas Geiping,
• Abstract summary: Traditional prompt-engineering approaches may prove ineffective once red-teaming turns into a weak-to-strong problem.<n>We evaluate more than 500 attacker-target pairs using LLM-based jailbreak attacks that mimic human red-teamers.<n>We derive a jailbreaking scaling law that predicts attack success for a fixed target based on attacker-target capability gap.
• Score: 71.89259138609965
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: As large language models grow in capability and agency, identifying vulnerabilities through red-teaming becomes vital for safe deployment. However, traditional prompt-engineering approaches may prove ineffective once red-teaming turns into a weak-to-strong problem, where target models surpass red-teamers in capabilities. To study this shift, we frame red-teaming through the lens of the capability gap between attacker and target. We evaluate more than 500 attacker-target pairs using LLM-based jailbreak attacks that mimic human red-teamers across diverse families, sizes, and capability levels. Three strong trends emerge: (i) more capable models are better attackers, (ii) attack success drops sharply once the target's capability exceeds the attacker's, and (iii) attack success rates correlate with high performance on social science splits of the MMLU-Pro benchmark. From these trends, we derive a jailbreaking scaling law that predicts attack success for a fixed target based on attacker-target capability gap. These findings suggest that fixed-capability attackers (e.g., humans) may become ineffective against future models, increasingly capable open-source models amplify risks for existing systems, and model providers must accurately measure and control models' persuasive and manipulative abilities to limit their effectiveness as attackers.

Related papers

• Quality-Diversity Red-Teaming: Automated Generation of High-Quality and Diverse Attackers for Large Language Models [34.601888589730194]
  This paper introduces Quality-Diversity Red-Teaming (QDRT), a new framework designed to address these limitations.<n>QDRT achieves goal-driven diversity through behavior-conditioned training and implements a behavioral replay buffer in an open-ended manner.<n>Our empirical evaluation demonstrates that QDRT generates attacks that are both more diverse and more effective against a wide range of target LLMs.
  arXiv  Detail & Related papers  (2025-06-08T13:07:41Z)
• One Surrogate to Fool Them All: Universal, Transferable, and Targeted Adversarial Attacks with CLIP [8.41221824218595]
  We present UnivIntruder, a novel attack framework that relies solely on a single, publicly available CLIP model and publicly available datasets.<n>Our experiments show that our approach achieves an Attack Success Rate (ASR) of up to 85% on ImageNet and over 99% on CIFAR-10, significantly outperforming existing transfer-based methods.
  arXiv  Detail & Related papers  (2025-05-26T11:25:00Z)
• Strategize Globally, Adapt Locally: A Multi-Turn Red Teaming Agent with Dual-Level Learning [39.931442440365444]
  AlgName is a novel red-teaming agent that emulates sophisticated human attackers through complementary learning dimensions.<n>AlgName enables the agent to identify new jailbreak tactics, develop a goal-based tactic selection framework, and refine prompt formulations for selected tactics.<n> Empirical evaluations on JailbreakBench demonstrate our framework's superior performance, achieving over 90% attack success rates against GPT-3.5-Turbo and Llama-3.1-70B within 5 conversation turns.
  arXiv  Detail & Related papers  (2025-04-02T01:06:19Z)
• Exploring the Adversarial Vulnerabilities of Vision-Language-Action Models in Robotics [68.36528819227641]
  This paper systematically quantifies the robustness of VLA-based robotic systems.<n>We introduce two untargeted attack objectives that leverage spatial foundations to destabilize robotic actions, and a targeted attack objective that manipulates the robotic trajectory.<n>We design an adversarial patch generation approach that places a small, colorful patch within the camera's view, effectively executing the attack in both digital and physical environments.
  arXiv  Detail & Related papers  (2024-11-18T01:52:20Z)
• SEAS: Self-Evolving Adversarial Safety Optimization for Large Language Models [19.486685336959482]
  Large language models (LLMs) continue to advance in capability and influence, ensuring their security and preventing harmful outputs has become crucial.<n>A promising approach to address these concerns involves training models to automatically generate adversarial prompts for red teaming.<n>We introduce the $mathbfStextelf-mathbfEtextvolving mathbfAtextdversarial mathbfStextafetyety mathbf(SEAS)$ optimization framework, which enhances security by leveraging data generated by the model itself.
  arXiv  Detail & Related papers  (2024-08-05T16:55:06Z)
• Learning diverse attacks on large language models for robust red-teaming and safety tuning [126.32539952157083]
  Red-teaming, or identifying prompts that elicit harmful responses, is a critical step in ensuring the safe deployment of large language models.<n>We show that even with explicit regularization to favor novelty and diversity, existing approaches suffer from mode collapse or fail to generate effective attacks.<n>We propose to use GFlowNet fine-tuning, followed by a secondary smoothing phase, to train the attacker model to generate diverse and effective attack prompts.
  arXiv  Detail & Related papers  (2024-05-28T19:16:17Z)
• Multi-granular Adversarial Attacks against Black-box Neural Ranking Models [111.58315434849047]
  We create high-quality adversarial examples by incorporating multi-granular perturbations. We transform the multi-granular attack into a sequential decision-making process. Our attack method surpasses prevailing baselines in both attack effectiveness and imperceptibility.
  arXiv  Detail & Related papers  (2024-04-02T02:08:29Z)
• MF-CLIP: Leveraging CLIP as Surrogate Models for No-box Adversarial Attacks [65.86360607693457]
  No-box attacks, where adversaries have no prior knowledge, remain relatively underexplored despite its practical relevance.<n>This work presents a systematic investigation into leveraging large-scale Vision-Language Models (VLMs) as surrogate models for executing no-box attacks.<n>Our theoretical and empirical analyses reveal a key limitation in the execution of no-box attacks stemming from insufficient discriminative capabilities for direct application of vanilla CLIP as a surrogate model.<n>We propose MF-CLIP: a novel framework that enhances CLIP's effectiveness as a surrogate model through margin-aware feature space optimization.
  arXiv  Detail & Related papers  (2023-07-13T08:10:48Z)
• Jailbroken: How Does LLM Safety Training Fail? [92.8748773632051]
  "jailbreak" attacks on early releases of ChatGPT elicit undesired behavior. We investigate why such attacks succeed and how they can be created. New attacks utilizing our failure modes succeed on every prompt in a collection of unsafe requests.
  arXiv  Detail & Related papers  (2023-07-05T17:58:10Z)
• Avoid Adversarial Adaption in Federated Learning by Multi-Metric Investigations [55.2480439325792]
  Federated Learning (FL) facilitates decentralized machine learning model training, preserving data privacy, lowering communication costs, and boosting model performance through diversified data sources. FL faces vulnerabilities such as poisoning attacks, undermining model integrity with both untargeted performance degradation and targeted backdoor attacks. We define a new notion of strong adaptive adversaries, capable of adapting to multiple objectives simultaneously. MESAS is the first defense robust against strong adaptive adversaries, effective in real-world data scenarios, with an average overhead of just 24.37 seconds.
  arXiv  Detail & Related papers  (2023-06-06T11:44:42Z)
• Boosting Black-Box Attack with Partially Transferred Conditional Adversarial Distribution [83.02632136860976]
  We study black-box adversarial attacks against deep neural networks (DNNs) We develop a novel mechanism of adversarial transferability, which is robust to the surrogate biases. Experiments on benchmark datasets and attacking against real-world API demonstrate the superior attack performance of the proposed method.
  arXiv  Detail & Related papers  (2020-06-15T16:45:27Z)

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.