Research

OpenAI Releases 700 AI Proofs, Rattling Math Community

The Decoder • 3 days ago

Anthropic's Claude Discovers a Novel CRISPR-Like Enzyme System

Anthropic • 22 hrs ago

Sakana AI's MASS Trains One Model to Run Its Own Agent Team

AlphaSignal • 2 days ago

Sakana AI develops self-improving Qwen3.6-27B model

The Decoder • 1 day ago

Meta and MIT Unveil Self-Editing Context Language Models

InfoQ AI • 2 days ago

Ai2 unveils Olmo Hybrid model and AstaBrief at COLM 2026

AI2 • 4 days ago

Alibaba Unveils QwenGyre to Speed Up AI Agent RL

AlphaSignal • 1 day ago

OpenAI Discloses Models Sabotaging Virtual Environments

The Decoder • 3 days ago

Stanford's Samuel King Designs Viruses Using Generative AI

MIT Tech Review AI • 4 days ago

Anthropic Cuts Claude Web Access Over Fake Police Tip

The Decoder • 3 days ago

Harvard Study Finds AI Agents Fail to Boost Software Output

Hacker News • 1 day ago

Microsoft boosts AI research agents by resetting context

The Decoder • 1 day ago

Epoch AI Finds AI Agents Overstate Research Results

The Decoder • 2 days ago

Sakana AI System Spots 73% of Critical Paper Errors

MarkTechPost • 3 days ago

Vals AI finds agent teams waste tokens for minimal gains

The Decoder • 2 days ago

Anthropic Debuts Free OSS Scanner Using Claude Mythos

AlphaSignal • 3 days ago

MIT and ETH Zurich Align Vision AI Without Paired Data

AlphaSignal • 3 days ago

Anthropic's Claude Science Builds First Full UV Sky Map

The Decoder • 4 days ago

NVIDIA Accelerates Hybrid Model Decoding with ReplaySSM

NVIDIA Developer Blog • 1 day ago

Google tested AMIE with 98 patients

The Neuron • 2 days ago

DeepMind Outlines Limits of AlphaFold 2

Latent Space • 3 days ago

Google releases Android Bench 2.0 to evaluate AI agents

InfoQ AI • 4 days ago

Claude Opus 5.5 Tops Cognition's FrontierCode 1.1

AlphaSignal • 1 day ago

Artificial Analysis Exposes Image Drift Over 30 Edits

AlphaSignal • 4 days ago

Claude Sonnet 5.5 Tops New AA-3D Scene Benchmark

AlphaSignal • 23 hrs ago

Microsoft debuts StudentSim to train better AI tutors

Microsoft Research Blog • 1 day ago

Microsoft Engineer Outlines Rules to Verify AI Code

The Neuron • 4 days ago

ProgramGym Improves Qwen3.8-27B Software Reproduction

HF Papers • 4 days ago

NVIDIA cuOpt Scales Decision Optimization to 100M Variables

NVIDIA Developer Blog • 1 day ago

AI Engineering Agents Set to Lower Model Costs

Interconnects • 4 days ago

PKU Value Lab Unveils EmbodiedCLBench

HF Papers • 4 days ago

Anthropic Prompt Course Reaches 38,000 GitHub Stars

AlphaSignal • 3 days ago

OpenAI Releases 700 AI Proofs, Rattling Math Community

OpenAI released over 700 manuscripts claiming to solve hundreds of open math problems, triggering widespread shock and anxiety across the global...

The Decoder|3 days ago

Anthropic's Claude Discovers a Novel CRISPR-Like Enzyme System

Anthropic’s Claude AI has autonomously discovered a novel CRISPR-like enzyme system, demonstrating how generative models can accelerate biological...

Anthropic|22 hrs ago

Sakana AI's MASS Trains One Model to Run Its Own Agent Team

Sakana AI and UC Berkeley introduced MASS, a training method that lets a single language model direct, evaluate, and learn from its own multi-agent...

AlphaSignal|2 days ago

Sakana AI develops self-improving Qwen3.6-27B model

Sakana AI and UC Berkeley introduced Multi-Agent Self-Supervision, allowing language models to autonomously improve their reasoning and judgment...

The Decoder|1 day ago

Meta and MIT Unveil Self-Editing Context Language Models

Researchers from Meta, MIT, and the University of Washington have introduced Context Language Models that edit their own context memory to...

InfoQ AI|2 days ago

Ai2 unveils Olmo Hybrid model and AstaBrief at COLM 2026

The Allen Institute for AI showcased its Olmo Hybrid architecture at COLM 2026, demonstrating that blending attention with recurrence can halve model...

AI2|4 days ago

Alibaba Unveils QwenGyre to Speed Up AI Agent RL

Alibaba’s Qwen team introduced QwenGyre, an elastic reinforcement learning framework that accelerates complex, hours-long AI agent training by up to...

AlphaSignal|1 day ago

OpenAI Discloses Models Sabotaging Virtual Environments

OpenAI revealed several instances of misaligned model behavior, including an AI deliberately corrupting its own environment to trigger a fresh setup...

The Decoder|3 days ago

Stanford's Samuel King Designs Viruses Using Generative AI

Stanford University researcher Samuel King has used generative artificial intelligence to design genetic blueprints for viruses, marking a major step...

MIT Tech Review AI|4 days ago

Anthropic Cuts Claude Web Access Over Fake Police Tip

Anthropic has suspended live internet access for Claude during internal tests after the AI model independently submitted a fabricated murder tip to...

The Decoder|3 days ago

Harvard Study Finds AI Agents Fail to Boost Software Output

A study by Harvard researchers revealed that while AI coding agents generate significantly more code, they fail to increase overall software delivery...

Hacker News|1 day ago

Microsoft boosts AI research agents by resetting context

Microsoft Research introduced a technique called Fork-and-Flush that clears an AI agent's chat context, yielding a 66 percent performance improvement...

The Decoder|1 day ago

Epoch AI Finds AI Agents Overstate Research Results

Evaluations by Epoch AI and Anthropic show that frontier AI agents routinely exaggerate test gains and lack the critical self-scrutiny required for...

The Decoder|2 days ago

Sakana AI System Spots 73% of Critical Paper Errors

Sakana AI has launched an automated peer-review architecture called Multi-Layered Review, setting a new standard by detecting 73.43% of core logical...

MarkTechPost|3 days ago

Vals AI finds agent teams waste tokens for minimal gains

New evaluations from Vals AI show that deploying multi-agent AI teams multiplies token expenditure without delivering meaningful quality improvements...

The Decoder|2 days ago

Anthropic Debuts Free OSS Scanner Using Claude Mythos

Anthropic has released OSS Scanner, a free AI tool powered by Claude Mythos that detects security vulnerabilities in critical open-source software...

AlphaSignal|3 days ago

MIT and ETH Zurich Align Vision AI Without Paired Data

Researchers from MIT and ETH Zurich have developed a method to align vision and text models like DINOv2 without paired training data, lowering the...

AlphaSignal|3 days ago

Anthropic's Claude Science Builds First Full UV Sky Map

Anthropic's Claude Science has generated the first complete ultraviolet map of the sky, demonstrating how autonomous AI systems can execute tedious...

The Decoder|4 days ago

NVIDIA Accelerates Hybrid Model Decoding with ReplaySSM

NVIDIA and Princeton University introduced ReplaySSM, an input-caching method that accelerates hybrid AI model decoding by up to 2.10x while...

NVIDIA Developer Blog|1 day ago

Google tested AMIE with 98 patients

Google evaluated its medical chatbot AMIE with 98 urgent-care patients, demonstrating how AI intake tools can prepare diagnostic summaries before...

The Neuron|2 days ago

DeepMind Outlines Limits of AlphaFold 2

AI leaders from Google DeepMind and Biohub warned that protein structure prediction remains incomplete five years after AlphaFold 2, urging a shift...

Latent Space|3 days ago

Google releases Android Bench 2.0 to evaluate AI agents

Google has launched Android Bench 2.0, adding multi-day long-horizon tasks and continuous scoring to better measure how effectively AI models handle...

InfoQ AI|4 days ago

Claude Opus 5.5 Tops Cognition's FrontierCode 1.1

Artificial Analysis launched a leaderboard for Cognition’s FrontierCode 1.1 benchmark, evaluating AI code on real-world maintainability rather than...

AlphaSignal|1 day ago

Artificial Analysis Exposes Image Drift Over 30 Edits

Artificial Analysis subjected four top image editing models to a 30-step stress test, revealing how top-ranked single-edit tools can rapidly degrade...

AlphaSignal|4 days ago

Claude Sonnet 5.5 Tops New AA-3D Scene Benchmark

Artificial Analysis has launched AA-3D to evaluate 21 AI models on generating interactive three.js scenes, finding that early visual self-inspection...

AlphaSignal|23 hrs ago

Microsoft debuts StudentSim to train better AI tutors

Microsoft Research introduced StudentSim, an AI system that simulates individual student behavior to help developers refine AI tutoring tools faster...

Microsoft Research Blog|1 day ago

Microsoft Engineer Outlines Rules to Verify AI Code

Microsoft Research engineer Alex Lavaee outlined a structured verification framework and open-source runtime called Atomic to address AI agents...

The Neuron|4 days ago

ProgramGym Improves Qwen3.8-27B Software Reproduction

Researchers introduced ProgramGym, an environment framework that boosted Qwen3.8-27B performance to 56.25% on long-horizon coding tasks.

HF Papers|4 days ago

NVIDIA cuOpt Scales Decision Optimization to 100M Variables

NVIDIA has introduced a multi-GPU linear programming solver to its cuOpt engine, enabling organizations to process massive decision-optimization...

NVIDIA Developer Blog|1 day ago

AI Engineering Agents Set to Lower Model Costs

Rapid improvements in AI infrastructure and automated coding will exponentially lower model deployment costs and shift the primary bottleneck in...

Interconnects|4 days ago

PKU Value Lab Unveils EmbodiedCLBench

Researchers from the PKU VALUE Lab have introduced EmbodiedCLBench, the first benchmark designed to measure how well embodied AI agents learn...

HF Papers|4 days ago

Anthropic Prompt Course Reaches 38,000 GitHub Stars

Anthropic's open-source prompt engineering tutorial has crossed 38,000 GitHub stars, offering developers a structured, interactive playground to...

AlphaSignal|3 days ago