Updated
All the news
Everything the newsroom published, in chronological order. Each item carries origin, sources and reading time.
1Today
4Yesterday
62This week
Wed, Jul 01
4 stories
RESEARCH Simple Prompting Baselines Outperform Complex Supervision Methods 
RESEARCH Researchers Close Gap Between AI Agents and Hand-Curated Skills 
RESEARCH New Training Technique Improves LLM Confidence Calibration by 63% 
RESEARCH Google Releases Zero-Shot Tabular Model but Hides Benchmark Data Mon, Jun 29
2 stories
RESEARCH Vision-language models route knowledge through just 2.5% of network RESEARCH AI Agents Double Repository-Level Merge Friction Sat, Jun 27
3 stories
RESEARCH Original-Language Context Recovers Accuracy Lost in Multilingual Cascades RESEARCH ENS Hits 10× Accuracy on Tough PDE Benchmarks Without Correction Loops 
RESEARCH Mechanism Taxonomy Lifts LLM Moderation F1 by 5.4% Fri, Jun 26
5 stories
RESEARCH Open-Weight Pipeline Achieves 68% Accuracy Extracting Political Networks from News 
RESEARCH Sequence Probability Fails as Production Inference Signal 
RESEARCH RiVER Enables Reinforcement Learning Without Ground-Truth Labels 
RESEARCH World Model Hallucination Is a Data Problem, Not Architecture 
RESEARCH Single Researcher Places 2nd in ICRA Robot-Folding Challenge Thu, Jun 25
6 stories
RESEARCH Models Shed Learned Rules During Training 
RESEARCH Free Scoring Signal Emerges from Standard RL Post-Training Runs 
RESEARCH DeepMind Forensic Protocol Diagnoses Confused vs. Misaligned AI 
RESEARCH Multimodal Models Flip Answers When Evidence Order Changes 
RESEARCH Production Voice AIs Ignore Emotion, Approving Fraud and Ending Care Calls 
RESEARCH Qwen's 397B Model Simulates Agent Environments Better Than GPT-5.4 Wed, Jun 24
4 stories
RESEARCH FFASR Benchmark Exposes Far-Field Speech Recognition Gap RESEARCH Strict Regex Fix Raises Agent Grading Recall by 60 Percentage Points 
RESEARCH InSight Enables Robots to Autonomously Learn New Tasks 
RESEARCH OpenThoughts-Agent Dataset Hits 44.8% on Agentic Benchmarks Tue, Jun 23
1 story
Mon, Jun 22
2 stories
RESEARCH Amortized In-Context Learning Cuts Few-Shot Serving Cost 
RESEARCH Google DeepMind's DiffusionGemma 28.6X harder to interpret than autoregressive models Fri, Jun 19
5 stories
RESEARCH Princeton Releases LOCUS, Machine-Readable Corpus of 9,239 US Local Ordinances 
RESEARCH Three Agents Beat Every SQL Benchmark With Zero Fine-Tuning 
RESEARCH OpenAnt LLM Pipeline Flags 28 Exploitable Vulnerabilities in OpenSSL RESEARCH MIT Extracts Attention Logic Into Swappable Python Code RESEARCH Physics-Augmented Koopman Networks Guarantee Generalization on Irregular Meshes Thu, Jun 18
1 story
Mon, Jun 15
5 stories
RESEARCH Single Dense Model Hosts Hundreds of Agent Personas as Lightweight Masks 
RESEARCH Sparse Attention Heads Redirect Vision-Language Models With 83% Accuracy 
RESEARCH ClinHallu Dissects Why Medical LLMs Misread Images 65% of the Time 
RESEARCH Component Interaction, Not Quality, Determines Agent Performance 
RESEARCH DiffusionGemma's Actual Decoding Contradicts Google's Block-Autoregressive Claims Sun, Jun 14
1 story
Sat, Jun 13
1 story
Load more stories Loading…