Kui Ren

Rubric Judgments Lack Uncertainty: Conformal Weighting for LLM Evaluation

Mitigating Cross-Modal Hallucinations by Correcting Modality Misrouting in Omnimodal LLMs

ASCENT: First-Order Optimal Fine-Tuning with Recalibration for Safety--Utility Co-Enhancement

Task-Aware Joint Pruning and Distillation for Efficient Audio Deepfake Detection

CollageAttack: Exploiting Cross-Modal Alignment Flaws in T2I Models through Spatial Text Composition

DeShortcut-Align: Decoupling Spurious Shortcuts for Robust Safety Alignment in Large Reasoning Models

EOPSA: Efficient On-Policy Self-Distilled Safety Alignment

Beyond Final Decisions: A Process-Centric Benchmark for Transparent AI-Assisted Peer Review

ActBench: Self-Evolving Benchmark of Behavioral Safety in Cowork Agents

From Role Prompt to Infinite Thinking: Exploiting Persona Conditioning for Inference Cost Attacks in LLMs

DataShield: Uncovering Risky Fine-Tuning Data Across LLMs Through Consensus Subspace Alignment

Ghosts Beneath Textures: Texture-Relation Cues for Cross-Paradigm AI-Generated Image Detection

ConsisGuard: Aligning Safety Deliberation with Policy Enforcement in LLM Guardrails

TRACE: Task-Aware Adaptive Self-Evolving Agentic Jailbreaking

RouteScan: A Non-Intrusive Approach to Auditing MoE LLMs Safety via Expert Routing Telemetry

"Training robust watermarking model may hurt authentication!'' Exploring and Mitigating the Identity Leakage in Robust Watermarking

LoopTrap: Termination Poisoning Attacks on LLM Agents

Defense against Poisoning Attacks under Shuffle-DP

When AI reviews science: Can we trust the referee?

Unveiling Deepfakes: A Frequency-Aware Triple Branch Network for Deepfake Detection

Channel-Level Semantic Perturbations: Unlearnable Examples for Diverse Training Paradigms

Towards Identification and Intervention of Safety-Critical Parameters in Large Language Models

FlowHijack: A Dynamics-Aware Backdoor Attack on Flow-Matching Vision-Language-Action Models

JANUS: A Lightweight Framework for Jailbreaking Text-to-Image Models via Distribution Optimization

STEP: Detecting Audio Backdoor Attacks via Stability-based Trigger Exposure Profiling

A Causal Perspective for Enhancing Jailbreak Attack and Defense

RerouteGuard: Understanding and Mitigating Adversarial Risks for LLM Routing

SpatialJB: How Text Distribution Art Becomes the "Jailbreak Key" for LLM Guardrails