FrameMorrow: Future-guided Frame Selection with Prospective Tokens for Long-Horizon Video Generation Paper • 2609.38839 • Published Sep 30 • 45
BadWAM: When World-Action Models Dream Right but Act Wrong Paper • 2607.15207 • Published Jul 16 • 42
On-Policy Self-Evolution via Failure Trajectories for Agentic Safety Alignment Paper • 2605.11882 • Published May 12 • 14
Visual Generation in the New Era: An Evolution from Atomic Mapping to Agentic World Modeling Paper • 2604.28185 • Published Apr 30 • 90
Vision-Language-Action Safety: Threats, Challenges, Evaluations, and Mechanisms Paper • 2604.23775 • Published Apr 26 • 44
AutoMIA: Improved Baselines for Membership Inference Attack via Agentic Self-Exploration Paper • 2604.01014 • Published Apr 1 • 10
Anatomy of a Lie: A Multi-Stage Diagnostic Framework for Tracing Hallucinations in Vision-Language Models Paper • 2603.15557 • Published Mar 16 • 30