When Refusal Looks Safe: The Refusal-Cue Shortcut in Safety Guard Models
Explorar
Noticias de IA
37834 elementos — filtrados, clasificados y sin duplicados
ProPRL: Property-Aware Prerequisite Relation Learning in Educational Knowledge Graphs
Learning a Vector-Symbolic Model for Socio-Cultural Tasks
CUADebug: Diagnosing and Repairing Computer-Use Agent Failures
BAP-SQL: Budget-Aware Observation Planning for Agentic Text-to-SQL
Balancing Efficiency and Efficacy: Training-Free Attention-Guided Switching Between Expli…
GAP-URGENet: A Generative-Predictive Fusion Framework for Universal Speech Enhancement
StuPASE: Towards Low-Hallucination Studio-Quality Generative Speech Enhancement
GUI-Lens: Coarse-to-Fine Cropping for GUI Grounding with General-Purpose VLMs
MDArena: Evaluating Coding Agents on Realistic Molecular Dynamics Workflows
Social Pressure Breaks Majority Voting in LLM Safety Panels
China’s Official Newspaper Warns US Over Expanding Tech Curbs
muSync-GS: Physics-Synchronized Driving Video Synthesis for Weather and Geometric Road Ha…
Training-Free Hashing-Based Attention via Binary Principal Components
Elbow-Based MoE Routing: A Training-Free Inference Time Plugin for Expert Selection
NOLLI: A Difficulty-Calibrated Puzzle Benchmark for Diagnosing the English-Korean Perform…
'Holistic' AI Safety Approach Needed: Rhodium's Goujon
EdgeLM: Edge Demonstrations for Language Models' Table Understanding
Image Classification Using CNN-QNN Hybrid Model with Optimized Correlated Features
SpaceX’s AI Splurge Puts a Damper on Debut Earnings
The Calibration Floor: Format Repair Can Masquerade as Self-Correction at Small-to-Mid Sc…
HyPASE: Hyperbolic Geometry for Parameter-Efficient Speech Emotion Fine-Tuning Framework …
Looking in the Mirror: Introspecting Side-Effect Misalignments Induced by Fine-Tuning
Chinese Optical Stocks Slide on Report of Proposed US Import Ban
[AINews] Megakernels are so dead and so back
Right Reset: Chunking by Prefix Removal
An Analysis and Implementation of Seam Carving for Content-Aware Image Resizing
Trident : How to Break Deep Reinforcement Learning Cyber Defenses (Agentic)
Adversarial Attacks for Good: A Survey of Proactive Protection across the Visual Content …
Pun Intended: Multi-Agent Translation of Wordplay with Contrastive Learning and Phonetic-…