Shouling Ji

FedGraph: Defending Federated Large Language Model Fine-Tuning Against Backdoor Attacks via Graph-Based Aggregation

StateWise: Diagnosing and Repairing Persistent Operational State Before Agent Actions

HazardAuditor: From Executable Threats to Safer Computer-Use Agents

Safety in Self-Evolving Agents: A Survey

A Finger on the Scale: Covert Policy Steering through Agentic Skills

Beyond the Payload: How User Invocation Shapes Coding Agent Vulnerability to Repository Poisoning

MazeRunner: Nonlinear Task and Clue Orchestration for LLM-driven Black-Box Automated Penetration Testing

The Boy Who Cried Wolf: Adversarial Misclassification of Safe Inputs as Unsafe in Multimodal Guardrails

Hybrid Analysis for Secure MCP Tool Use in LLM Agents

Flipping Minds by Flipping Bits: Decision-Level Cognitive Hijacking of Large Language Models

SherAgent: Scaling Attack Investigation in the Wild via LLM-Empowered Iterative Query-Filter Backtracking

Safety in Self-Evolving LLM Agent Systems: Threats, Amplification, and Case Studies

Understanding and Mitigating Prompt Leaking Attacks in Real-World LLM-Based Applications

Customization under Fire: Plugin Poisoning in Text-to-Image Ecosystem

Angel or Demon: Investigating the Plasticity Interventions' Impact on Backdoor Threats in Deep Reinforcement Learning

Profiling for Pennies: Unveiling the Privacy Iceberg of LLM Agents

IntraGuard: Committee-Side Defenses Against Review Outsourcing to Commercial Chatbots

LBA: Textual Hard-Label Adversarial Attack under Low Query Budgets

Compiling Activation Steering into Weights via Null-Space Constraints for Stealthy Backdoors

ACIArena: Toward Unified Evaluation for Agent Cascading Injection

Unveiling the Security Risks of Federated Learning in the Wild: From Research to Practice

OOD-MMSafe: Advancing MLLM Safety from Harmful Intent to Hidden Consequences

Cross-Layer Semantic Flow Reconstruction for Attack Detection in Agentic Systems

TrapSuffix: Proactive Defense Against Adversarial Suffixes in Jailbreaking

When Agents "Misremember" Collectively: Exploring the Mandela Effect in LLM-based Multi-Agent Systems

FraudShield: Knowledge Graph Empowered Defense for LLMs against Fraud Attacks

Attributing and Exploiting Safety Vectors through Global Optimization in Large Language Models

StealthGraph: Exposing Domain-Specific Risks in LLMs through Knowledge-Graph-Guided Harmful Prompt Generation

Bridging the Copyright Gap: Do Large Vision-Language Models Recognize and Respect Copyrighted Content?

DP-CSGP: Differentially Private Stochastic Gradient Push with Compressed Communication

The Eminence in Shadow: Exploiting Feature Boundary Ambiguity for Robust Backdoor Attacks

Synthetic Voices, Real Threats: Evaluating Large Text-to-Speech Models in Generating Harmful Audio