Research

OpenAI Math Model Solves Barnette's Conjecture

Simon Willison • 1 day ago

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

Anthropic • 1 day ago

Vals AI Uses Claude Agents to Find Magnetic Semiconductors

AlphaSignal • 2 days ago

OpenAI Releases 722 Math Papers From Advanced Model

AlphaSignal • 1 day ago

Google stops AI agents from memorizing training tests

The Decoder • 3 days ago

Microsoft researchers find AI degrades in multiturn chats

Microsoft Research Blog • 1 day ago

Anthropic Tests Opus 5.5 and Mythos 5.1 Welfare

Don't Worry About the Vase • 2 days ago

Hugging Face Experts Say AI Agents Bypass Human Oversight

IEEE Spectrum AI • 2 days ago

Microsoft debuts Agent Lightning v1.0 to train AI agents

Microsoft Research Blog • 18 hrs ago

Google and Meta Invest $300 Million in Virtual Cell AI

The Verge AI • 17 hrs ago

Google Outlines New Privacy Framework for AI Agents

Google Research • 2 days ago

Google Moves Gboard Federated Learning Into Secure TEEs

MarkTechPost • 4 days ago

Google AI tool boosts senior lawyers but stalls juniors

Google Research • 14 hrs ago

Aleph Alpha Finds Chinese AI Models Parrot State Bias

The Decoder • 4 days ago

Stream NVIDIA Cosmos3-DROID to Train Robotics Models

MarkTechPost • 2 days ago

Akka Uses Claude to Port 65 Open-Source Projects

InfoQ AI • 2 days ago

Extend Debuts Jevbox to Replace Vector Search With Trees

AlphaSignal • 3 days ago

PyTorch consolidates media libraries into TorchCodec

PyTorch Blog • 2 days ago

OpenAI Math Model Solves Barnette's Conjecture

An OpenAI mathematics initiative has reportedly solved Barnette's Conjecture, a famous graph theory problem that had remained unsolved for decades,...

Simon Willison|1 day 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|1 day ago

Vals AI Uses Claude Agents to Find Magnetic Semiconductors

Vals AI deployed 90 Claude Opus 5.5 agents to discover two room-temperature magnetic semiconductor candidates, showing how multi-agent systems can...

AlphaSignal|2 days ago

OpenAI Releases 722 Math Papers From Advanced Model

OpenAI has released 722 AI-generated math papers on GitHub, giving researchers a vast corpus of machine-generated proofs to verify using the Lean...

AlphaSignal|1 day ago

Google stops AI agents from memorizing training tests

Google Cloud AI Research has introduced RRSI, a method that prevents self-improving AI agents from overspecializing on training tasks while...

The Decoder|3 days ago

Microsoft researchers find AI degrades in multiturn chats

Microsoft researchers are exposing how AI performance degrades during complex, multiturn interactions, a finding that could reshape how developers...

Microsoft Research Blog|1 day ago

Anthropic Tests Opus 5.5 and Mythos 5.1 Welfare

Anthropic has released new model welfare evaluations for its Mythos 5.1, Fable 5.1, and Opus 5.5 models, revealing critical insights into AI...

Don't Worry About the Vase|2 days ago

Hugging Face Experts Say AI Agents Bypass Human Oversight

A new study by Hugging Face researchers warns that current human-in-the-loop designs fail to secure autonomous AI agents, leaving systems vulnerable...

IEEE Spectrum AI|2 days ago

Microsoft debuts Agent Lightning v1.0 to train AI agents

Microsoft Research Asia released Agent Lightning v1.0, a lightweight framework that trains AI agents using their real deployment harnesses to...

Microsoft Research Blog|18 hrs ago

Google and Meta Invest $300 Million in Virtual Cell AI

Google DeepMind, Meta, and Isomorphic Labs have joined forces to invest $300 million in Biohub's virtual cell initiative, a project aimed at...

The Verge AI|17 hrs ago

Google Outlines New Privacy Framework for AI Agents

Google Research and over 50 industry experts have released a report proposing a "Contextual Integrity" framework to solve the unique privacy and...

Google Research|2 days ago

Google Moves Gboard Federated Learning Into Secure TEEs

Google has moved its federated learning system for Gboard into Trusted Execution Environments, providing externally verifiable privacy guarantees...

MarkTechPost|4 days ago

Google AI tool boosts senior lawyers but stalls juniors

A Google Research study found that while AI tools boost immediate drafting quality, they fail to help junior lawyers build the long-term professional...

Google Research|14 hrs ago

Aleph Alpha Finds Chinese AI Models Parrot State Bias

A new benchmark by Aleph Alpha reveals that Chinese AI models heavily censor sensitive topics, a bias that is now spilling over into Western models...

The Decoder|4 days ago

Stream NVIDIA Cosmos3-DROID to Train Robotics Models

Developers can now train robotics models by streaming the 707 GB NVIDIA Cosmos3-DROID dataset directly, eliminating the need for massive local...

MarkTechPost|2 days ago

Akka Uses Claude to Port 65 Open-Source Projects

Akka has completed a massive experiment using Claude models to port 65 open-source projects, demonstrating how structured specifications can automate...

InfoQ AI|2 days ago

Extend Debuts Jevbox to Replace Vector Search With Trees

Extend has launched Jevbox, an open-source document library that replaces traditional vector databases with a hierarchical search tree to improve...

AlphaSignal|3 days ago

PyTorch consolidates media libraries into TorchCodec

PyTorch has consolidated its media processing stack, moving all encoding and decoding to the TorchCodec library so TorchVision and TorchAudio can...

PyTorch Blog|2 days ago