November 2025 — page 1 of 2
- Study reveals reasoning AI models often skip real thinking during self-checks· AI
- Branching Networks Learn Multiple Algorithms Faster and More Efficiently· AI
- LLMs Can Build Better Knowledge Bases for Temporal Reasoning—If Schema Design Comes First· AI Research
- GAPS Framework Combines Static Analysis With Dynamic Testing to Pinpoint Android App Methods· Software Engineering
- New Framework Shields AI Agent Networks from Language-Based Attacks· AI
- Researchers compress lengthy AI system prompts into a single learned token· AI
- New Multimodal Translation Model Combines Audio, Images, and Text Across Languages· AI
- Dripper: Lightweight AI Model Extracts Web Content Fast and Accurately· AI
- New synthetic benchmark tool tackles evaluation gaps in causal machine learning· AI
- New Dataset and Training Method Boost Medical AI's Ability to Reason Across Multiple Images· AI
- New Clustering Method Tackles Mixed Data Types in Big Data Era· Machine Learning
- LLMs Struggle to Generalize Across Task Difficulties, New Study Shows· AI
- LLMs Show Promise for Test Generation But Face Key Quality Gaps· AI
- Efficient Fix for Stale Code Advice: New Technique Helps AI Models Recommend Current APIs· AI
- AI models can generate and apply their own safety rules to resist manipulation attempts· AI
- Researchers Cut Audio Token Load in Large Language Models by Three-Fold· AI
- AlphaFold Drives Faster Discovery Across Biology After Five Years of Impact· AI
- Self-Supervised Pre-training Enables Learning from Noisy Labels Without Clean Data· Machine Learning
- Researchers Propose SORL to Stabilize AI Agent Training with Multi-Turn Interactions· AI
- Study Uncovers Disconnect Between Understanding and Image Generation in AI Models· AI
- New Framework Uses AI to Discover Better Market Signals at Scale· AI
- LLM Safety Training Backfires: New Method Fixes Over-Refusal While Keeping Systems Secure· AI
- ThreadWeaver Brings Parallel Reasoning to Language Models Without Accuracy Loss· AI
- How Familiar Concepts Make AI Models More Stable Under Challenge· AI
- Researchers Introduce FSU-QA Benchmark to Test AI Models' Ability to Anticipate Future Events· AI
- Small Language Models Can Fix Code Errors in Domain-Specific Languages Using Reinforcement Learning· AI
- New method efficiently explores alternative interpretable AI models with identical accuracy· AI
- Code Summaries Improve LLM Bug Repair, but Modest Gains Persist· AI
- New Framework Optimizes LLM Training by Combining Fine-Tuning with Bias Correction· Agentic AI
- New AI system synchronizes dubbed speech with video using visual cues and speech synthesis· AI
- Multi-Agent AI Framework Automates Software Logging Decisions at 96% Accuracy· AI
- New AI Framework Combines Multiple Deep Learning Models for Bangla News Classification· AI
- Stronger CLIP Teachers Don't Always Produce Better Students, Study Finds· AI
- Multi-agent system harnesses LLMs to automate REST API testing at scale· AI
- New hybrid method tackles missing data in single-cell RNA analysis· AI
- Researchers Use AI to Simulate Civic Meetings with Speaker-Specific Personas· AI
- New unsupervised learning approach detects network attacks without labeled training data· AI
- Scientists Lag Behind Latest AI Models, Study Finds· AI
- Hybrid System Predicts Both Common and Rare Behaviors on Bluesky Social Network· AI
- LLMs demand new evaluation methods for music recommendation systems· AI
- AI Model Predicts Heart Failure Patient Risk at Discharge to Guide Care Planning· AI, Healthcare
- Google DeepMind releases Nano Banana Pro, advancing AI image generation with studio-quality output· AI
- Vision Models Learn to Handle Mismatched Tasks with Text-Driven Prompting· AI
- AnyBox Achieves 36-Point Leap in Zero-Shot Box Detection for Warehouse Robots· Robotics
- Multi-Agent LLMs Dramatically Improve Incident Response Quality Over Single-Agent Systems· AI
- New Method Enables Foundation Models to Learn Domain Tasks Without Forgetting Core Abilities· AI
- Researchers Use Formal Verification to Test Multi-Robot Systems Safely Before Deployment· Robotics
- Researchers Launch "Project Rachel," an AI Author Pose to Test Academic Publishing· AI
- Google DeepMind unveils Gemini 3, its most advanced AI model yet· AI
- Bayesian inference explains why people accept nonsensical sentences· AI
- New benchmark measures how well AI models convert written instructions into software diagrams· AI
- New steganographic backdoor technique bypasses existing safeguards against AI model poisoning· AI Security
- New Method Teaches AI Models to Correct Math Mistakes Like a Teacher Would· AI
- Large-Scale Study Reveals Security Gaps in Agent Context Files for AI Coding Tools· Agentic AI
- New tool uses AI to automatically update security exploits across library versions· AI
- New Model-Blending Technique Boosts LLM Performance Without Expensive Retraining· AI
- Researchers develop framework to identify and manage accountability gaps in AI-powered software· AI
- A Hidden Weakness in Topic Modeling: Consistent Results Don't Guarantee Accuracy· AI
- New Framework Gives AI Agents Memory Across Sessions to Learn from Web Browsing Mistakes· Agentic AI
- Researchers mine 56,000 reviews to decode what users really think about AI· AI
- New AI Network Advances Detection of Hidden Objects in Natural Environments· AI
- eFLINT: A Domain-Specific Language for Automating Legal Compliance in Software· AI & Programming Languages
- New Framework Tackles LLM Code Errors by Cross-Checking Against Problem Variants· AI
- New pipeline combines extractive and abstractive methods for legal document summarization· AI
- New AI Method Detects Violence in CCTV Footage with 95% Accuracy Using Short Video Clips· AI
- Study reveals how AI coding agents struggle with multi-hunk bug repairs· Agentic AI
- New Framework Uses Hyperbolic Geometry to Identify Subtle Visual Differences in Fine-Grained Classification· AI
- Researchers Expose Vulnerabilities in Bengali Language AI Models and Propose Defense Strategy· AI Security
- Hybrid Vision Transformer and Frequency Analysis Method Sharpens Blurry Images· AI
- New routing system directs AI models to their strongest coding and software tasks· AI
- Researchers Develop Efficient Method to Attack Visual AI Agents Without Model Access· AI Security
- Hybrid AI System Brings Formal Verification to Natural Language Policies· AI
- New Benchmark Reveals How Vision-Language Models Struggle to Turn Descriptions Into Working Crochet Patterns· AI
- AI Chatbot Trained for Mental Health Shows Promise in Year-Long Pilot Study· AI
- π-Attention Brings Linear-Cost Long-Context Transformers to Practical Workloads· AI
- New Framework Assigns Credit and Blame to Individual LLMs in Multi-Agent Systems· AI
- AI Framework Automatically Writes Security Vulnerability Detectors at 53% Accuracy· Security & AI
- New Technique Uses Selective Reprocessing to Boost LLM Reasoning Accuracy· AI
- Researchers Scale Humanoid Robots to 42M Parameters, Achieving Natural Whole-Body Movement· Robotics
- Researchers propose "Logits-to-Logic" framework to fix LLMs' logical inconsistencies in structured knowledge tasks· AI
- Local AI Models Now Match Cloud Performance While Using Far Less Power· AI
- Researchers Build AI Framework to Automatically Generate Equity Research Reports from Financial Data· AI
- New approach lets voice agents respond instantly while tapping powerful reasoning models· AI
- Implicit Simulation Speeds Soft Robot Learning by Up to 40x· Robotics
- Researchers Build Verified System to Prevent AI from Making False Scientific Discoveries· AI
- AI Disclosure Standards in Ethics Research Need Process Transparency, Not Just Reporting· AI
- Study reveals how spurious correlations in training data cause language models to hallucinate with confidence· AI
- Machine learning spots language patterns that signal depression severity· Healthcare AI
- Modular Sparse Experts Enable LLMs to Learn Continuously Without Forgetting Old Knowledge· AI
- Using LLMs and representative agents to scale realistic traffic models· AI
- New Rust-Based Tool Accelerates Grammar-Constrained Test Input Generation by Thousands of Times· Software Engineering
- Multi-agent system learns to generate personalized math questions with controlled variety· Agentic AI
- New Method Removes Unsafe Content from AI Image and Video Models Without Quality Loss· AI Safety
- Fast model screening for wearable activity trackers cuts training time by 99%· AI
- New Benchmark Reveals How Far AI Falls Short at Optimizing Real-World Code Performance· AI
- New Framework Reveals How Gender Stereotypes Skew LLM Reasoning in Logic Puzzles· AI
- Vision-Language AI Model Improves Early Alzheimer's Detection Through Handwriting Analysis· AI
- New AI Model Detects Cognitive Distortions in Text for Mental Health Support· AI
- Study links Stack Overflow improvements to real open-source code fixes· Software Engineering
- New watermarking technique harnesses AI to embed invisible markers for detecting deepfakes· AI