Adaptive Instruction Composition uses a neural contextual bandit with RL to adaptively combine crowdsourced texts, generating more effective and diverse LLM jailbreaks than random or prior adaptive methods on Harmbench.
MART: improving LLM safety with multi- round automatic red-teaming
9 Pith papers cite this work, alongside 4 external citations. Polarity classification is still indexing.
9
Pith papers citing it
4
external citations · external index
citation-role summary
background 2
method 1
citation-polarity summary
polarities
background 3representative citing papers
CREST-Search is a red-teaming framework that crafts seemingly benign search queries to induce unsafe citations from web-augmented LLMs, backed by a new WebSearch-Harm dataset for fine-tuning a specialized attacker model.
Transient Turn Injection is a new attack that evades LLM moderation by spreading harmful intent over multiple isolated turns using automated agents.
Changing the internal reasoning structure of large reasoning models through simple supervised fine-tuning on 1K examples produces strong safety alignment that generalizes across tasks and languages.
Iterative self-improving codebooks enhance safety in autoregressive multimodal models by self-identifying unsafe generations and updating the codebook to eliminate harmful visual token mappings without external feedback.
A multi-role red-teaming framework with attacker, target, and jury LLMs measures faithfulness in English and Arabic, finding false-premise prompts and length limits change unfaithfulness rates.
The paper releases a 1,554-prompt consensus-labeled bank separating executable malicious code requests from security knowledge requests, validated by five-model majority labeling with Fleiss' kappa of 0.876.
The paper introduces a taxonomy of AI safety for LLMs organized into Trustworthy AI, Responsible AI, and Safe AI perspectives, accompanied by a review of state-of-the-art methods, challenges, and future directions.
A survey that creates taxonomies for jailbreak attacks and defenses on LLMs, subdivides them into sub-classes, and compares evaluation approaches.
citing papers explorer
-
Adaptive Instruction Composition for Automated LLM Red-Teaming
Adaptive Instruction Composition uses a neural contextual bandit with RL to adaptively combine crowdsourced texts, generating more effective and diverse LLM jailbreaks than random or prior adaptive methods on Harmbench.
-
When Search Goes Wrong: Red-Teaming Web-Augmented Large Language Models
CREST-Search is a red-teaming framework that crafts seemingly benign search queries to induce unsafe citations from web-augmented LLMs, backed by a new WebSearch-Harm dataset for fine-tuning a specialized attacker model.
-
Transient Turn Injection: Exposing Stateless Multi-Turn Vulnerabilities in Large Language Models
Transient Turn Injection is a new attack that evades LLM moderation by spreading harmful intent over multiple isolated turns using automated agents.
-
Reasoning Structure Matters for Safety Alignment of Reasoning Models
Changing the internal reasoning structure of large reasoning models through simple supervised fine-tuning on 1K examples produces strong safety alignment that generalizes across tasks and languages.
-
Safe Autoregressive Image Generation with Iterative Self-Improving Codebooks
Iterative self-improving codebooks enhance safety in autoregressive multimodal models by self-identifying unsafe generations and updating the codebook to eliminate harmful visual token mappings without external feedback.
-
A Red Teaming Framework for Large Language Models: A Case Study on Faithfulness Evaluation
A multi-role red-teaming framework with attacker, target, and jury LLMs measures faithfulness in English and Arabic, finding false-premise prompts and length limits change unfaithfulness rates.
-
A Validated Prompt Bank for Malicious Code Generation: Separating Executable Weapons from Security Knowledge in 1,554 Consensus-Labeled Prompts
The paper releases a 1,554-prompt consensus-labeled bank separating executable malicious code requests from security knowledge requests, validated by five-model majority labeling with Fleiss' kappa of 0.876.
-
AI Safety Landscape for Large Language Models: Taxonomy, State-of-the-art, and Future Directions
The paper introduces a taxonomy of AI safety for LLMs organized into Trustworthy AI, Responsible AI, and Safe AI perspectives, accompanied by a review of state-of-the-art methods, challenges, and future directions.
-
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.