Expert evaluation of jailbreak responses is costly and difficult to scale, so the community increasingly relies on automated evaluators to determine whether an attack succeeds. However, jailbreak studies typically validate their chosen evaluator independently, repeatedly spending resources on similar evaluation efforts while making results across papers difficult to compare. Different evaluators also encode different definitions of jailbreak success, meaning that reported attack strength and apparent progress can depend substantially on which evaluator is used. We systematically compare six evaluators that recur in recent jailbreak attack and defense research: HarmBench, JailbreakBench, JailbreakRadar, StrongReject, JADES, and JailMeter. To our knowledge, no prior study has evaluated all six on the same human-labeled data under a controlled setup. We evaluate them on JailbreakQR and JailMeter-Eva, using human judgments as the reference, and measure agreement with humans, error types, and consistency across attack families. For evaluators that require a general-purpose LLM judge, we use a shared backbone to control for model-specific variation. We found that JADES exhibits the best overall performance, while HarmBench and StrongReject also demonstrate good performance.
Evaluating the effectiveness of multi-turn jailbreak attacks on large language models (LLMs) increasingly relies on automated evaluators, yet their reliability and agreement with human judgments in cybersecurity conversations have not been systematically examined. This paper presents a systematic human-machine agreemen...
Michael Tchuindjang, Nathan Duran, Phil Legg et al.· International Symposium on N...· 0 citations
This work proposes Sequential Epistemic and Action-Level Validation (SEAV), a verification-centric jailbreak evaluation framework that decomposes responses into ordered steps and evaluates both validity and correctness, and shows that enforcing correctness substantially reshapes measured robustness.
Qilong Wu, Sahil Wadhwa, Pranab Mohanty et al.· 0 citations
Internal safety scores judge a prompt before any text is generated, and they are validated by how well they separate harmful prompts from benign ones, so Active Attention Probing is introduced, which supplies a fixed content independent measurement coordinate.
Ming-Yu Luo, Ming Deng, Zilang Qiu et al.· 0 citations
It is found that user hostility varies 13-fold across models, driven largely by who each model attracts rather than by model behaviour: first-turn hostility spreads far wider than post-response hostility, and more than fifteenfold separates the extremes even after deduplicating opening prompts.
Fan-Qi Zeng, Sadid A. Hasan, Chao-Cheng He· Proceedings of the 10th Work...· 0 citations
LLM judges are inherently subjective, often favoring different responses in pairwise comparison when neither option is objectively wrong. To study this subjectivity, we introduce JudgeProfile, a framework that dissects LLM evaluation into perception (how a judge compares two responses across specific attributes like cl...
Qiong-Qiong Cao, Kang-Ni Liu, Xuan Kan et al.· 0 citations
LLJ Cards is introduced, a framework that synthesizes best practices from measurement theory, natural language generation, and machine learning literature into practical guidelines for LLJ-based evaluations and provides a structured framework for applying these principles in the design and reporting of automated evalua...
Research Manager Mike Walker reflects on a career spanning hacker culture, cybersecurity, and AI, sharing lessons about trust, teamwork, leadership, and why solving difficult problems requires curiosity, judgment, agility—and an understanding of the rules. The post Learn the rules: The hacker mindset appeared first on Microsoft Research.
MIT News · Artificial Intelligence· news.mit.eduNov 17, 2026
By studying the computational foundations of language learning, Associate Professor Jacob Andreas builds intelligent systems that can communicate effectively with humans.
Microsoft Research Blog· microsoft.comNov 16, 2026
AI coding agents are increasingly help developers fix bugs and build features largely on their own, but every interaction with their models consumes tokens. In each session, an agent sends instructions, tool descriptions, and session history to its model. Most efforts to reduce this cost focus on individual tasks. Yet developers return to the same […] The post TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient appeared first on Microsoft Research.
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.