July 2025 — page 1 of 2
- New AI Model Uses Latent Space Learning to Generate Synthetic Time Series Data· Machine Learning
- New framework trains medical AI to better combine knowledge retrieval with clinical reasoning· Agentic AI
- New AI Framework Uses Neural Features Instead of Words to Discover Hidden Topics in Text· AI
- Why large language models struggle with Bengali: tokenizer design, not script complexity· AI
- New benchmark reveals leading AI models execute computer-control attacks at dangerously high rates· AI Safety
- Researchers Deploy First Validated Algorithm to Detect Rank Reversal in Decision-Making Systems· AI
- Researchers Propose New Formal Framework for Measuring Intelligence· AI
- Multi-Agent AI System Tackles Credit Assessment With Hierarchical LLM Framework· AI
- Google DeepMind's AlphaEarth Foundations Maps the Planet Using AI and Satellite Data· AI
- LLMs trained to reason can spot fake bug fixes that look correct on paper· AI
- Quantum algorithms break classical limits in reinforcement learning with simulator access· Quantum AI
- New Framework Optimizes Bug Detection by Intelligently Managing Multiple Fuzzing Tools· Software Engineering
- New Framework Treats LLM Agent Memory Like Git, Boosting Long-Horizon Task Performance· Agentic AI
- New benchmark reveals how AI models struggle with accurate harm reduction guidance for drug users· AI
- New Framework Brings Mathematical Transparency to Deep Learning for Wireless Sensing· AI
- How Adjectives Shift What Sounds Plausible: New Study Tests Models on Semantic Change· AI
- LLMs Learn to Reason Better by Training on Their Own Uncertainty· AI
- New framework helps researchers design AI prompts for systematic content analysis· AI
- New GPU-Accelerated Benchmark Tackles Multi-Agent Robot Learning for Home Assistance· AI
- Researchers Bridge Neural and Symbolic AI With New Framework for Graph Data· AI
- LLMs Show Promise as Choice Modelling Assistants, But Performance Varies Sharply· AI
- New Method Generates Tests Aligned With Real Development Intentions· Software Engineering
- New Method Uses Expert Annotations to Improve Treatment Decisions From Patient Data· AI
- LLMs tackle null-pointer bugs at scale with NullRepair system· AI
- Researchers Trace How Chain-of-Thought Reasoning Works Inside AI Models· AI
- New reasoning framework helps AI understand complex social situations by breaking them into three distinct cognitive stages· AI
- LLM-based framework CIgrate dramatically outperforms rule-based approach for automating CI service migration· AI
- Deep learning model sharpens low-resolution turbulence simulations threefold faster than traditional methods· Machine Learning
- Federated learning framework bridges synthetic-to-real gap in autonomous driving vision· Agentic AI
- Eloquence team tackles multilingual speech recognition with layered projector approach· AI
- Researchers Use AI to Auto-Generate Compiler Fuzzing Tools, Uncovering Dozens of New Bugs· AI
- CodeEvo: AI agents collaborate to generate high-quality code training data· AI
- Survey Maps How Game Theory Models Agents With Conflicting Beliefs and Perceptions· AI
- Trump's "Anti-Woke" AI Order Masks Government Bias, Critics Say· AI
- New Benchmark Tests AI Models on Multiple Languages and Media Types· AI
- New steering technique guides AI model behavior without retraining· AI
- Synthetic images enable faster, more accurate zero-shot domain adaptation without target data· AI
- Researchers map MITRE ATT&CK techniques to software supply chain risk management tasks· Cybersecurity
- New Benchmark Reveals AI Agents Struggle to Implement Features from Documentation· AI
- Study reveals widespread gaps in GitHub Actions workflow security and best practices· Software Engineering
- New AI Framework Automates 3D Software Design from Natural Language· AI
- New decoder technique speeds up LLM inference by extracting hidden predictions mid-computation· AI
- Researchers Develop Methods to Detect Copyrighted Code in AI Models· AI
- LLMs Guide Secure UAV Communications with AI-Powered Trajectory Planning· AI
- LLMs reveal why opposition parties attack more during campaigns across Europe· AI
- DeepMind's Aeneas Uses AI to Unlock Ancient Roman Inscriptions· AI
- LLM-Based Framework Disentangles Messy Code Commits with Collaborative Reasoning· AI
- Training Smaller LLMs to Generate Verified Code With Less Human Input· AI
- Synthetic Data Boosts Multilingual Agricultural AI for Farmers Across India· AI
- New benchmark measures AI safety risks in high-stakes professions· AI
- New metrics go beyond task success to measure robotic confidence and execution quality· AI
- Gemini Deep Think achieves gold-medal standard at International Mathematical Olympiad· AI
- New AI Framework Uses Satellite Imagery to Map Disaster Damage in Real Time· AI
- New benchmark reveals how well AI language models bridge knowledge across fields· AI
- Neuromorphic AI chip efficiently diagnoses skin lesions on pocket-sized devices· AI
- Robots Learn Better by Following Human Gaze, Not Analyzing Everything Equally· Robotics
- Survey Maps Vector Quantization as Key Bridge Between Continuous Data and Language Models· AI
- AI-powered robot handles post-surgery patient check-ins with adaptive conversation· AI
- New Method Uses Retrieval of Past Demonstrations to Boost Offline Reinforcement Learning· AI
- Showing Multiple AI Image Descriptions Helps Blind Users Spot Errors· AI Accessibility
- Research Tool Captures Fine-Grained Changes in Jupyter Notebooks to Advance Data Science Development Practices· Research
- Large language models show mixed ability to inject factual commonsense into logical reasoning· AI
- Study finds no gender gap in AI code assistant effectiveness, but reveals cognitive benefits· AI
- New AI approach tackles credit card fraud detection by rethinking how to balance imbalanced data· AI
- AI system with built-in safety checks tackles quantum chemistry simulations reliably· Agentic AI
- New Framework Uses Cognitive Comparison to Pick Best Model for Sentiment Analysis· AI
- Researchers Release Open Diabetes Datasets to Accelerate AI Development· AI
- New framework links information theory and statistical mechanics to understand model complexity trade-offs· AI
- New Framework Detects Synthetic User Activity in Cybersecurity Simulations· Cybersecurity
- Researchers develop personalized AI rewards through learned user summaries· AI
- AI models separate harmfulness detection from refusal responses, researchers discover· AI Safety
- New Benchmark Reveals How Far AI Falls Short at Real-World Code Optimization· AI
- New Framework Simulates Student Code Revisions to Test AI Tutors Before Deployment· AI
- Framework optimizes language model training by learning which task combinations work best together· AI
- Sparse Autoencoders Reveal Hidden Language-Specific Patterns in Multilingual AI Models· AI
- Meta-Learning Framework Helps Models Catch Code Problems They've Never Seen Before· AI
- New Method Uses AI Policy Learning to Optimize LLM Training on Edge Devices· AI
- Researchers Develop Precise Local Editing Method for 3D Gaussian Scenes· AI
- New benchmark tackles contamination crisis in AI code-fixing evaluation· Agentic AI
- Researchers Rethink Reinforcement Learning Through an Evolutionary Lens· Artificial Intelligence
- New Framework Shows How AI Can Serve Communities With Limited Data· AI
- Researchers test LLMs to rewrite abusive text while preserving intent· AI
- Evolutionary Algorithm Makes LLM Compression Faster and More Effective· AI
- New method grounds AI model explanations in solid mathematical theory· AI
- New framework maps MARL's gap between theory and real-world deployment· Artificial Intelligence
- New Method Tunes Language Models to Better Understand Financial Text Across Languages· AI
- NLP transforms insurance pricing with alternative data and text-derived risk insights· AI
- LLMs naturally organize emotions in hierarchies that mirror human psychology, but show biased recognition across demographics· AI
- AI Resume Screening Tools May Hide Incompetence Behind Appearance of Fairness· AI
- LLMs Show Promise for Tax Law Reasoning, But Aren't Ready for Full Automation· AI
- Researchers Crack the Logic Behind Two-Layer Networks with Smooth Activation Functions· AI
- AI agents with cryptocurrency access pose unprecedented security risks, researchers warn· AI Security
- New Method Achieves True Machine Unlearning by Erasing Features at the Source· Machine Learning
- Kimina-Prover-72B Achieves State-of-the-Art Theorem Proving with Test-Time RL Search· AI
- New Benchmark Tests How Well AI Agents Juggle Farm Work and Social Life in Virtual Worlds· Agentic AI
- New technique uses randomized keys to stop attackers from forging AI-generated watermarks· AI Security
- New Framework Makes Neural Network Pruning Scale-Invariant Using Projective Geometry· AI
- AI agents automate protein simulations, cutting setup time and human error· Agentic AI
- How LLMs process uncertainty remains largely hidden during inference· AI
- Deep learning model reaches 93.76% accuracy recognizing seven emotions from speech across five major datasets· AI