Xingjun Ma

WithAnyone: Toward Controllable and ID Consistent Image Generation

Toward Universal and Transferable Jailbreak Attacks on Vision-Language Models

RedTopic: Toward Topic-Diverse Red Teaming of Large Language Models

ADMIT: Few-shot Knowledge Poisoning Attacks on RAG-based Fact Checking

Unraveling Hallucination in Large Reasoning Models: A Topological Perspective

AgenticPA: Toward Automated and Large-Scale Prompt Attacks on LLMs

SafeEvalAgent: Toward Agentic and Self-Evolving Safety Evaluation of LLMs

What Do VLMs See? Benchmarking Vision-Language Models on Ambiguous Images

AutoBackdoor: Automating Backdoor Attacks via LLM Agents

Imperceptible Jailbreaking against Large Language Models

FreezeVLA: Action-Freezing Attacks against Vision-Language-Action Models

OmniSVG: A Unified Scalable Vector Graphics Generation Model

X-Transfer Attacks: Towards Super Transferable Adversarial Attacks on CLIP

SAMA: Towards Multi-Turn Referential Grounded Video Chat with Large Language Models

JailBound: Jailbreaking Internal Safety Boundaries of Vision-Language Models

Detecting Backdoor Samples in Contrastive Language Image Pretraining

Attention-based Graph Coreset Labeling for Active Learning

BlueSuffix: Reinforced Blue Teaming for Vision-Language Models Against Jailbreak Attacks

HOGT: High-Order Graph Transformers

AnyAttack: Self-supervised Generation of Targeted Adversarial Attacks for Vision-Language Models

Identity Lock: Locking API Fine-tuned LLMs With Identity-based Wake Words

TUAP: Targeted Universal Adversarial Perturbations for CLIP

IDEATOR: Jailbreaking VLMs Using VLMs

Towards a Theoretical Understanding of Memorization in Diffusion Models