mlsanityData to 7 Oct 2026

Revisiting the Robust Alignment of Circuit Breakers

arXiv 2407.15902 ↗First posted 22 Jul 2024cs.CRcited by 10

Not on the map. Why?

Abstract (arXiv)

Over the past decade, adversarial training has emerged as one of the few reliable methods for enhancing model robustness against adversarial attacks [Szegedy et al., 2014, Madry et al., 2018, Xhonneux et al., 2024], while many alternative approaches have failed to withstand rigorous subsequent evaluations. Recently, an alternative defense mechanism, namely "circuit breakers" [Zou et al., 2024], has shown promising results for aligning LLMs. In this report, we show that the robustness claims of "Improving Alignment and Robustness with Circuit Breakers" against unconstraint continuous attacks in the embedding space of the input tokens may be overestimated [Zou et al., 2024]. Specifically, we demonstrate that by implementing a few simple changes to embedding space attacks [Schwinn et al., 2024a,b], we achieve 100% attack success rate (ASR) against circuit breaker models. Without conducting any further hyperparameter tuning, these adjustments increase the ASR by more than 80% compared to the original evaluation. Code is accessible at: https://github.com/SchwinnL/circuit-breakers-eval

Key facts. Cited by 10 papers in the map (data through 7 Oct 2026). First posted on arXiv 22 Jul 2024.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 1; 2025: 6; 2026: 3.

10 papers.

  1. 2024LLM Defenses Are Not Robust to Multi-Turn Human Jailbreaks Yetcited by 145
  2. 2025FORTRESS: Frontier Risk Evaluation for National Security and Public Safetycited by 20
  3. 2025RL-Obfuscation: Can Language Models Learn to Evade Latent-Space Monitors?cited by 19
  4. 2025REINFORCE Adversarial Attacks on Large Language Models: An Adaptive, Distributional, and Semantic Objectivecited by 15
  5. 2025Jailbreaking to Jailbreakcited by 8
  6. 2025Adversarial Alignment for LLMs Requires Simpler, Reproducible, and More Measurable Objectivescited by 8
  7. 2025Improving Large Language Model Safety with Contrastive Representation Learningcited by 7
  8. 2026Large Language Models Generate Harmful Responses Using a Distinct Mechanism, Shared Across Harm Typescited by 3
  9. 2026TamperBench: Systematically Stress-Testing LLM Safety Under Fine-Tuning and Tamperingcited by 1
  10. 2026Reference-Grafting Matches Fine-Tuning at Eliciting Sandbagged Capabilitiescited by 1

Similar, not cited

Close in wording, no citation either way

  1. 2024Improving Alignment and Robustness with Circuit Breakers
  2. 2025Targeting Alignment: Extracting Safety Classifiers of Aligned LLMs
  3. 2025Replicating TEMPEST at Scale: Multi-Turn Adversarial Attacks Against Trillion-Parameter Frontier Models
  4. 2023Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM
  5. 2025Probing the Robustness of Large Language Models Safety to Latent Perturbations
  6. 2025AdversariaL attacK sAfety aLIgnment(ALKALI): Safeguarding LLMs through GRACE: Geometric Representation-Aware Contrastive Enhancement- Introducing Adversarial Vulnerability Quality Index (AVQI)
  7. 2026All you need to break LLMs are Black-Box, Adapting, Efficient, Transferable, Harmful, Applicable ... Attacks
  8. 2023Baseline Defenses for Adversarial Attacks Against Aligned Language Models