SOTAVerified

Red Teaming

Papers

Recently AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 101–150 of 251 papers

TitleStatusHype
SeqAR: Jailbreak LLMs with Sequential Auto-Generated CharactersCode0
Sowing the Wind, Reaping the Whirlwind: The Impact of Editing Language ModelsCode0
Stealthy and Persistent Unalignment on Large Language Models via Backdoor InjectionsCode0
Steering Without Side Effects: Improving Post-Deployment Control of Language ModelsCode0
Distract Large Language Models for Automatic Jailbreak AttackCode0
SAGE: A Generic Framework for LLM Safety EvaluationCode0
The Structural Safety Generalization ProblemCode0
TRIDENT: Enhancing Large Language Model Safety with Tri-Dimensional Diversified Red-Teaming Data SynthesisCode0
We Should Identify and Mitigate Third-Party Safety Risks in MCP-Powered Agent SystemsCode0
What Is Wrong with My Model? Identifying Systematic Problems with Semantic Data SlicingCode0
Red Teaming with Mind Reading: White-Box Adversarial Policies Against RL AgentsCode0
Investigating Bias Representations in Llama 2 Chat via Activation Steering—0
IterAlign: Iterative Constitutional Alignment of Large Language Models—0
JAB: Joint Adversarial Prompting and Belief Augmentation—0
DeceptPrompt: Exploiting LLM-driven Code Generation via Adversarial Natural Language Instructions—0
Jailbreaking GPT-4V via Self-Adversarial Attacks with System Prompts—0
Jailbreaking Large Language Models Against Moderation Guardrails via Cipher Characters—0
Jailbreaking Large Language Models with Symbolic Mathematics—0
Jailbreaking Multimodal Large Language Models via Shuffle Inconsistency—0
CURE: Concept Unlearning via Orthogonal Representation Editing in Diffusion Models—0
CulturalTeaming: AI-Assisted Interactive Red-Teaming for Challenging LLMs' (Lack of) Multicultural Knowledge—0
JBFuzz: Jailbreaking LLMs Efficiently and Effectively Using Fuzzing—0
KDA: A Knowledge-Distilled Attacker for Generating Diverse Prompts to Jailbreak LLMs—0
Know Thy Judge: On the Robustness Meta-Evaluation of LLM Safety Judges—0
LARGO: Latent Adversarial Reflection through Gradient Optimization for Jailbreaking LLMs—0
CTI4AI: Threat Intelligence Generation and Sharing after Red Teaming AI Models—0
CoT Red-Handed: Stress Testing Chain-of-Thought Monitoring—0
Conversational Complexity for Assessing Risk in Large Language Models—0
Learning from Red Teaming: Gender Bias Provocation and Mitigation in Large Language Models—0
Lessons From Red Teaming 100 Generative AI Products—0
Leveraging Reinforcement Learning in Red Teaming for Advanced Ransomware Attack Simulations—0
LLM-Assisted Red Teaming of Diffusion Models through "Failures Are Fated, But Can Be Faded"—0
Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red Teaming—0
LLM-Safety Evaluations Lack Robustness—0
LLMStinger: Jailbreaking LLMs using RL fine-tuned LLMs—0
The Human Factor in AI Red Teaming: Perspectives from Social and Collaborative Computing—0
LoRA Fine-tuning Efficiently Undoes Safety Training in Llama 2-Chat 70B—0
Low-Resource Languages Jailbreak GPT-4—0
MAD-MAX: Modular And Diverse Malicious Attack MiXtures for Automated LLM Red Teaming—0
Making Every Step Effective: Jailbreaking Large Vision-Language Models Through Hierarchical KV Equalization—0
MART: Improving LLM Safety with Multi-round Automatic Red-Teaming—0
Computational Red Teaming in a Sudoku Solving Context: Neural Network Based Skill Representation and Acquisition—0
MMDT: Decoding the Trustworthiness and Safety of Multimodal Foundation Models—0
Model Card and Evaluations for Claude Models—0
CELL your Model: Contrastive Explanations for Large Language Models—0
Multi-lingual Multi-turn Automated Red Teaming for LLMs—0
The Multilingual Alignment Prism: Aligning Global and Local Preferences to Reduce Harm—0
Can Large Language Models Change User Preference Adversarially?—0
Can Large Language Models Automatically Jailbreak GPT-4V?—0
Offensive Security for AI Systems: Concepts, Practices, and Applications—0
Show:102550
← PrevPage 3 of 6Next →

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1SUDOAttack Success Rate41—Unverified