August 2025 — page 1 of 2
- Legal AI System Tackles Phantom Citations, Lifts Citation Accuracy in Bar Exam Test· Agentic AI
- New Framework Prioritizes Human Connection Over Automation in Software Requirements Engineering· Software Engineering
- AI-Generated Explanations Still Fail to Teach Humans Effective Learning Strategies, Study Finds· AI
- New Framework Helps AI Systems Understand Hidden Social Dynamics· AI
- AI-Powered Tool Makes Threat Modeling Accessible to Resource-Constrained Organizations· AI Security
- Study exposes unreliability in LLM-generated robot control code· Robotics & AI
- AI Framework Protects Industrial Machine Tools From Cyber Attacks Using Adaptive Watermarking· AI Security
- New Method Selects Best Reasoning from Multiple AI Answers Using Confidence Scoring· AI
- New Framework Blocks AI Agents From Acting on Incomplete Instructions· AI
- New dataset reveals how LLMs fall for conspiracy theory framing, even when they reject misinformation· AI
- New Framework Tackles Over-Forgetting in AI Model Privacy Removal· AI
- New AI Method Uses Diffusion Models to Learn Brain Patterns from EEG Data· AI
- New Benchmark Reveals Multimodal AI Models Struggle With Basic Visual Perception Tasks· AI
- Researchers Propose Defense Against Malicious AI Model Training via Reinforcement Learning· AI Safety
- New AI System Detects and Corrects Pronunciation Errors in Quranic Recitation· AI
- LLM-Powered Multi-Agent Framework Automates Quantum Circuit Programming· AI
- Transformer Model Cuts Duplicate Crash Reports by 15%+ in Developer Triage Systems· AI in Software Engineering
- Real-Time Model Checking Lets Robots Plan Several Steps Ahead to Avoid Obstacles· Robotics
- New transformer network tackles precision roof mapping from aerial LiDAR data· AI
- Researchers Map How LLMs Organize Intelligence Like Animal Brains· AI
- New Survey Maps the Landscape of LLM-Based Agentic Reasoning Across Multiple Domains· Agentic AI
- Transformer-Based Robot Policy Learns Multiple Motor Tasks With One Model· Robotics
- AI Model Detects Suicide Risk in Teen Social Media Posts Before Crisis Occurs· AI
- New Neural Framework Tackles Missing Data in Time Series Analysis· AI
- New Framework Teaches AI to Learn Multimodal Data Through Relation Semantics· AI
- Survey Maps the Rise of Step-by-Step Visual Reasoning in AI Systems· AI
- Researchers Develop AI System That Learns Individual Driving Preferences· AI
- New AI system teaches humanoid robots to carry objects through cluttered spaces without stopping· Robotics
- New Dataset and AI Model Bring Advanced Reasoning to Eye Disease Diagnosis· AI
- New Framework Tackles Bias in Multimodal AI by Embracing Data Diversity· AI
- New defense system uses retrieval and examples to stop jailbreak attacks on AI models· AI Security
- New coverage metric measures metamorphic testing effectiveness more accurately than traditional code coverage· Software Engineering
- Chatbot Helps Engineers Master Essence Framework Through AI-Powered Coaching· AI
- LLM-assisted approach bridges semantic gaps in collaborative systems engineering with SysML v2· AI
- New Approach Achieves Superior Few-Shot Segmentation with 1/24th the Parameters· AI
- Getting L0 Wrong Breaks Sparse Autoencoders' Feature Learning, Researchers Find· AI
- How confident should LLMs be when generating code? A study in uncertainty· AI
- EyeMulator brings human code-reading patterns into AI language models· AI
- New training approach improves AI code completion by fixing unrealistic data patterns· AI
- Automated Agent System Slashes Time to Build LLM Safety Benchmarks from Months to Days· AI
- Tool Use Can Backfire in LLMs When Internal Reasoning Is Already Strong· AI
- MedRepBench: New Benchmark Measures Medical Report Understanding in AI Vision Models· AI
- New modular AI system improves sign language translation by splitting gesture, fingerspelling, and lip-reading tasks· AI
- New method helps AI models give more honest confidence estimates about their answers· AI
- New benchmark challenges AI systems to reason across scientific documents and images· AI
- Researchers develop autonomous drone swarm for wildlife tracking without central control· Robotics
- Retrieval beats loading everything: RAG outperforms long-context for EHR clinical tasks· AI
- Researchers Tackle Cross-Image Confusion in Multi-Image AI Models· AI
- New Vietnamese benchmark exposes vision-language model weaknesses beyond OCR· AI
- New Framework Combines Inverse Reinforcement Learning and Search to Boost LLM Reasoning Performance· AI
- New Framework Boosts LLMs' Understanding of Code Logic and Structure· AI
- Researchers map cognitive framework for adaptive LLM reasoning across three dimensions· AI
- Yokai Learning Environment challenges dominant zero-shot coordination benchmark with dynamic belief tracking· Agentic AI
- New AI System Builds Evolving Medical Knowledge Graph from Decades of Research· AI
- AI Agent Recovers Missing Links Between Software Issues and Code Commits· AI
- Researchers Flip Multimodal AI Design: Text-to-Vision Mapping Eliminates Need for Alignment Pretraining· AI
- New Mamba-Based AI Model Shows Promise for Detecting Liver Disease in MRI Scans· AI
- Researchers demonstrate exact data deletion in billion-parameter language models· AI
- Large language models spontaneously develop physics intuition without explicit training· AI
- SupraTok Tokenizer Breaks Whitespace Boundaries to Boost Language Model Efficiency· AI
- New Framework Compresses LLM Prompts Without Training to Cut Costs and Speed· AI
- New Benchmark Reveals LLMs Translate Code Correctly but Inefficiently· AI
- Researchers Develop Lightweight Framework for Parallel Robot Vision Tasks on Edge Hardware· Robotics
- New Approach Reproduces Mobile App Performance Problems Using User Reviews· Software Engineering
- Causal abstraction offers new framework for understanding how neural networks compute and explain behavior· AI
- New semi-supervised model handles multi-view data with incomplete views and missing labels· Machine Learning
- New benchmark reveals major gaps in AI vision models' ability to reason about object properties· AI
- Psyche-R1: New AI Model Blends Reasoning, Empathy for Mental Health Support· AI
- Researchers Build Dataset to Train AI Models on Analog Circuit Design· AI
- Study reveals how AI models pick up anti-homeless bias from social media and city meetings· AI
- AI-MO Releases Kimina-Prover-RL, Open-Source Pipeline for Lean 4 Theorem Proving· AI
- Why LLMs Struggle With Cross-Lingual Knowledge: New Research Points to Representation Unification· AI
- New KV Cache Compression Method Cuts Memory Use 32x While Matching Performance· AI
- New Framework Helps AI Designers Read Human Intent Through Text, Sketches, and Level Design· Agentic AI
- New Method Cuts Training Cost for AI Language Models by Reusing Internal Layers· AI
- New technique tunes individual neurons to balance LLM safety and usefulness· AI
- LoRA-Based Method Improves Pronunciation Control in Multilingual Speech Synthesis· AI
- New Method Uses Counterfactual Explanations to Spot Out-of-Distribution Data· AI
- Arm & ExecuTorch 0.7 bring generative AI to older smartphones and edge devices· AI
- New Framework Learns Hidden User Preferences to Build Smarter Mobile Agents· Agentic AI
- New Benchmark Reveals Major Gaps in How AI Models Follow Complex Logic Instructions· AI
- New technique reveals how language models process cultural knowledge internally· AI
- Multi-Agent Reinforcement Learning Reduces Strain of Tool Use in AI Systems· Agentic AI
- New synthetic training framework helps language models master agent-like reasoning tasks· Agentic AI
- New Framework Lets Researchers Inject, Detect, and Remove Biases from AI Language Models· AI
- Arm Launches Neural Super Sampling for Mobile Gaming and Graphics· AI
- New Training Method Helps Transformers Solve Harder Planning Problems· AI
- New VGGSounder Dataset Fixes Gaps in Audio-Visual Foundation Model Testing· AI
- New Platform Lets Kids Build AI-Powered AR Experiences Instead of Just Consuming Them· AI
- Machine Learning Selects Optimal Maximum Clique Solvers with 90% Accuracy· Machine Learning
- New Visual Tool Maps the Structure of Human-AI Conversations· AI
- AI Detection Tools Fail to Distinguish Human-AI Writing Collaborations in Education· AI
- Survey reveals wide gaps in AI ethics practices across development roles and countries· AI
- Hiding Unreliable AI Predictions Backfires: Clinicians Miss More Diagnoses When AI Stays Silent· AI
- New Neural Bridge Processes Strengthen Conditional Learning in Uncertain Data· AI
- AI System Uses Multi-Step Reasoning and Self-Checking to Improve Endoscopic Diagnosis· Agentic AI
- New Tool Helps Developers Make Sense of Bug-Finding Warnings Through Smart Summaries· Software Engineering
- New Framework Helps Language Models Forecast Time Series by Bridging Data Modality Gap· AI
- Encrypted Music Search: New Method Protects Embeddings Without Sacrificing Speed· AI
- Active inference framework models decision-making in neural networks mimicking biological brains· AI