VoxSafeBench reveals that speech language models recognize social norms from text but fail to apply them when acoustic cues like speaker or scene determine the appropriate response.
hub
arXiv preprint arXiv:2311.03348 (2023)
15 Pith papers cite this work. Polarity classification is still indexing.
15
Pith papers citing it
hub tools
citation-role summary
background 3
method 1
citation-polarity summary
polarities
background 4representative citing papers
Greedy random search recovers token sequences that elicit harmful response prefixes from LLMs without meaningful instructions, showing natural backdoors are present yet require more effort than semantic attacks.
Agreeableness in AI personas reliably predicts sycophantic behavior in 9 of 13 tested language models.
GAMBIT constructs gamified instructional traps that decompose harmful visuals and drive MLLMs to reconstruct and answer malicious queries as part of winning a game, achieving over 85% attack success on models including GPT-4o and Gemini 2.5 Flash.
Applies MAP-Elites quality-diversity optimization to evolve semantic attack strategies across dimensions like strategy type, encoding, and length, uncovering distinct vulnerability profiles in four LLMs including GPT-4o-mini and Claude 3.5 Sonnet.
PIA achieves lower attack success rates on persona-based jailbreaks via self-play co-evolution of attacks (PLE) and defenses (PICL) that structurally decouples safety from persona context using unilateral KL-divergence.
Conflicts between alignment objectives or dilemmas increase attack success rates on LRMs by shifting and overlapping safety and functional neural representations.
TRACE-RPS drops LLM attribute inference accuracy from around 50% to below 5% via fine-grained anonymization plus a two-stage rejection optimization.
JailbreakBench supplies an evolving set of jailbreak prompts, a 100-behavior dataset aligned with usage policies, a standardized evaluation framework, and a leaderboard to enable comparable assessments of attacks and defenses on LLMs.
StrongREJECT provides a standardized benchmark and evaluator for jailbreak attacks that aligns better with human judgments than prior methods and reveals that successful jailbreaks often reduce model capabilities.
PAIR uses an attacker LLM to iteratively craft effective jailbreak prompts for black-box target LLMs in fewer than 20 queries.
Proposes a probabilistic framework for latent agentic substructures in DNNs using log-score utilities and log pooling, with proofs on unanimity and an application to persona emergence in LLM alignment.
GUARD automates generation of guideline-violating questions and jailbreak diagnostics to test LLM compliance with government ethics guidelines, validated empirically on eight models and extended to vision-language models.
A survey that creates taxonomies for jailbreak attacks and defenses on LLMs, subdivides them into sub-classes, and compares evaluation approaches.
Impersonating complex misaligned personas via biographies and role-play bypasses safety in ChatGPT, Gemini, and Deepseek, succeeding on 38-40 out of 40 illicit questions across tested models.
citing papers explorer
-
VoxSafeBench: Not Just What Is Said, but Who, How, and Where
VoxSafeBench reveals that speech language models recognize social norms from text but fail to apply them when acoustic cues like speaker or scene determine the appropriate response.
-
On the Hardness of Junking LLMs
Greedy random search recovers token sequences that elicit harmful response prefixes from LLMs without meaningful instructions, showing natural backdoors are present yet require more effort than semantic attacks.
-
Too Nice to Tell the Truth: Quantifying Agreeableness-Driven Sycophancy in Role-Playing Language Models
Agreeableness in AI personas reliably predicts sycophantic behavior in 9 of 13 tested language models.
-
GAMBIT: A Gamified Jailbreak Framework for Multimodal Large Language Models
GAMBIT constructs gamified instructional traps that decompose harmful visuals and drive MLLMs to reconstruct and answer malicious queries as part of winning a game, achieving over 85% attack success on models including GPT-4o and Gemini 2.5 Flash.
-
Quality-Diversity Evolution for Discovering Diverse Vulnerabilities in LLM Safety
Applies MAP-Elites quality-diversity optimization to evolve semantic attack strategies across dimensions like strategy type, encoding, and length, uncovering distinct vulnerability profiles in four LLMs including GPT-4o-mini and Claude 3.5 Sonnet.
-
Disentangling Intent from Role: Adversarial Self-Play for Persona-Invariant Safety Alignment
PIA achieves lower attack success rates on persona-based jailbreaks via self-play co-evolution of attacks (PLE) and defenses (PICL) that structurally decouples safety from persona context using unilateral KL-divergence.
-
Conflicts Make Large Reasoning Models Vulnerable to Attacks
Conflicts between alignment objectives or dilemmas increase attack success rates on LRMs by shifting and overlapping safety and functional neural representations.
-
Stop Tracking Me! Proactive Defense Against Attribute Inference Attack in LLMs
TRACE-RPS drops LLM attribute inference accuracy from around 50% to below 5% via fine-grained anonymization plus a two-stage rejection optimization.
-
JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models
JailbreakBench supplies an evolving set of jailbreak prompts, a 100-behavior dataset aligned with usage policies, a standardized evaluation framework, and a leaderboard to enable comparable assessments of attacks and defenses on LLMs.
-
A StrongREJECT for Empty Jailbreaks
StrongREJECT provides a standardized benchmark and evaluator for jailbreak attacks that aligns better with human judgments than prior methods and reveals that successful jailbreaks often reduce model capabilities.
-
Jailbreaking Black Box Large Language Models in Twenty Queries
PAIR uses an attacker LLM to iteratively craft effective jailbreak prompts for black-box target LLMs in fewer than 20 queries.
-
Probabilistic Modeling of Latent Agentic Substructures in Deep Neural Networks
Proposes a probabilistic framework for latent agentic substructures in DNNs using log-score utilities and log pooling, with proofs on unanimity and an application to persona emergence in LLM alignment.
-
GUARD: Guideline Upholding Test through Adaptive Role-play and Jailbreak Diagnostics for LLMs
GUARD automates generation of guideline-violating questions and jailbreak diagnostics to test LLM compliance with government ethics guidelines, validated empirically on eight models and extended to vision-language models.
-
Jailbreak Attacks and Defenses Against Large Language Models: A Survey
A survey that creates taxonomies for jailbreak attacks and defenses on LLMs, subdivides them into sub-classes, and compares evaluation approaches.
-
Dr. Jekyll and Mr. Hyde: Two Faces of LLMs
Impersonating complex misaligned personas via biographies and role-play bypasses safety in ChatGPT, Gemini, and Deepseek, succeeding on 38-40 out of 40 illicit questions across tested models.