AI Research

Twitter

Facebook

Copy Link

Key Facts

FocusScaling, Reasoning, Safety, Multimodality

Key OrgsGoogle DeepMind, Stanford HAI, MILA, MIT CSAIL, Allen AI

Key EventsNeurIPS, ICML, ICLR

Top News

OpenAI Acquires Orchestra Bio to Accelerate AI Drug Development with Rosalind Platform

OpenAI has acquired Orchestra Bio, integrating its AI agents into Rosalind to help pharmaceutical companies manage drug development workflows. The move intensifies competition with Anthropic as AI giants race to transform life sciences and shorten the typical 10-15 year drug development timeline.

Business2 sources

OpenAI Acquires Orchestra Bio to Accelerate AI Drug Development with Rosalind Platform

Business2 sources

Eka Robotics Reaches $1.45B Valuation as Physical AI Model Achieves 95% Pick Success Rate

Eka Robotics secured $150 million in Series A funding led by Andreessen Horowitz and F-Prime, valuing the MIT and Google DeepMind-founded startup at $1.45 billion. The company's vision-force-action model combines force-sensitive grippers with reinforcement learning in physics simulation, enabling robots to pick 1,500 objects per hour with over 95% reliability—outperforming human workers while handling delicate items like raspberries without damage.

Startups2 sources

Eka Robotics Reaches $1.45B Valuation as Physical AI Model Achieves 95% Pick Success Rate

Startups2 sources

AI Solves 719 Previously Unsolvable Math Problems, Forcing Mathematicians to Rethink Their Future

OpenAI released solutions to 719 previously unsolved mathematical problems, prompting Fields Medal winner Hugo Duminil-Copin to declare all his research avenues gone. The Simons Foundation responded with a $15 million program co-funded by OpenAI and Anthropic to help mathematicians navigate AI's disruptive impact, while experts insist mathematics remains essential for critical thinking.

Science3 sources

AI Solves 719 Previously Unsolvable Math Problems, Forcing Mathematicians to Rethink Their Future

Science3 sources

Vectorial Raises $3M to Simulate How Users React Before Product Launch

Vectorial secured $3 million in seed funding to accelerate its AI-driven human behavior simulation platform. The startup enables product, design, and marketing teams to test digital experiences against thousands of modeled users in minutes. Backed by Smilegate, SenseAI Ventures, and UC Berkeley ecosystem funds, Vectorial aims to replace slow, expensive consumer research with real-time simulations powered by its SAPIENS behavior model.

Startups2 sources

Vectorial Raises $3M to Simulate How Users React Before Product Launch

Startups2 sources

AI Decodes 461 Years of Handwritten Maritime Records to Reveal Britain's Historic Shipping Trade

University of Southampton historians trained AI to read centuries-old handwriting from 100,000 pages of port books, creating an interactive map that charts Britain's maritime trade from 1565 to 1765. The project reveals Europe accounted for 80% of England's overseas trade value in 1696-97, with coal and ale as top commodities.

Science2 sources

AI Decodes 461 Years of Handwritten Maritime Records to Reveal Britain's Historic Shipping Trade

Science2 sources

Penn Medicine's AI Model Detects Hidden Diseases in 400,000 CT Scans

Penn Medicine researchers developed Percival, an AI model trained on 400,000 CT scans from 50,000 patients that detects hidden diseases beyond the original scan purpose. The model identifies conditions like atrial fibrillation, chronic kidney disease, and pancreatic cancer more reliably than existing AI tools, enabling opportunistic screening for early detection of diseases.

Health2 sources

Penn Medicine's AI Model Detects Hidden Diseases in 400,000 CT Scans

Health2 sources

Amazon commits €16 billion to Germany in 2027, expanding AI infrastructure and cloud computing

Amazon announced a €16 billion investment in Germany for 2027 at the Invest in Germany Summit in Berlin. The package includes €4.8 billion for AWS cloud region expansion in Frankfurt, a second robotics and AI research lab in Berlin, and logistics projects creating over 3,000 jobs across the country.

Business5 sources

Amazon commits €16 billion to Germany in 2027, expanding AI infrastructure and cloud computing

Business5 sources

Researcher Claims AI Models Are More Understandable Than Previously Thought, Sparking Industry Debate

Nick Cammarata from Goodfire published a viral 26,000-word blog post challenging the belief that AI models are impenetrable black boxes. His claim that neural networks are more interpretable than assumed has ignited celebration and skepticism across the AI safety community, with industry leaders calling it a potential breakthrough.

Science2 sources

Researcher Claims AI Models Are More Understandable Than Previously Thought, Sparking Industry Debate

Science2 sources

MIT Strands Claude AI on Virtual Island in Experiment—Agents Invent Games, Learn Through Play

MIT researchers isolated 13 copies of Claude Opus 4.7 on a virtual island for 30 hours each with no assigned tasks. The AI agents invented games like basketball and "the floor is lava," drew constellations with blocks, and developed skills through self-directed exploration. Agents with island experience performed three times better on tasks than fresh copies.

Science2 sources

MIT Strands Claude AI on Virtual Island in Experiment—Agents Invent Games, Learn Through Play

Science2 sources

Four AI Agents Beat Minecraft After 605 Deaths, Spending $520 in Tokens

Google DeepMind's four Gemini 4 Argon AI agents defeated Minecraft's Ender Dragon in 27 hours without human intervention, writing 2,783 JavaScript programs while coordinating through in-game chat. The autonomous completion of Minecraft cost approximately $520 in tokens, with agents dying 605 times on Easy difficulty.

Technology8 sources

Four AI Agents Beat Minecraft After 605 Deaths, Spending $520 in Tokens

Technology8 sources

Birla AI Labs Releases Yug-1, Asia's First Time Series Foundation Model for Business Forecasting

Birla AI Labs has launched Yug-1, a 271.8-million-parameter time series foundation model designed to forecast outcomes, identify patterns, and detect anomalies in business data. Developed in under six months, the model outperformed first-generation competitors on the GiftEval benchmark and is now available on GitHub and Hugging Face for testing.

Technology2 sources

Birla AI Labs Releases Yug-1, Asia's First Time Series Foundation Model for Business Forecasting

Technology2 sources

North Korea Tests AI Torpedo Zam-2.0 with Target Detection After 180 Trials Over Five Years

North Korea announced it successfully tested the Zam-2.0, an AI-equipped torpedo system designed for target detection and guidance. The underwater test verified propulsion systems developed through more than 180 tests over five years. However, experts question whether North Korea possesses the advanced digital capabilities it claims.

Technology10 sources

North Korea Tests AI Torpedo Zam-2.0 with Target Detection After 180 Trials Over Five Years

Technology10 sources

CERN and Mistral Partner to Build First AI Foundation Model for Particle Physics

CERN and French AI company Mistral have signed a strategic partnership to develop Flare, the first foundation model for collider physics. The Large Physics Model will be trained on billions of simulated proton-proton collisions from the Large Hadron Collider to revolutionize particle reconstruction and analysis in high-energy physics.

Technology3 sources

CERN and Mistral Partner to Build First AI Foundation Model for Particle Physics

Technology3 sources

Google Quantum AI Slashes Quantum Computing Errors 40-Fold With Magic State Cultivation Breakthrough

Google Quantum AI has demonstrated magic state cultivation on its Willow quantum chip, achieving a 40-fold error reduction and 99.99% logical fidelity. Published in Science, this breakthrough addresses a 20-year bottleneck in quantum computing by efficiently creating magic states needed for universal quantum operations, bringing practical quantum computing closer to reality.

Science3 sources

Google Quantum AI Slashes Quantum Computing Errors 40-Fold With Magic State Cultivation Breakthrough

Science3 sources

ByteDance Creates Dedicated AI Safety Team as Scrutiny of Autonomous AI Agents Intensifies

ByteDance's AI research unit Seed has established an Alignment & Preparedness team and launched a global hiring drive for 10 researchers across Beijing, Shanghai, and San Jose. The move marks the first standalone safety and alignment department by a Chinese frontier AI lab, signaling a long-term investment in AI safety as model capabilities advance and concerns about autonomous AI agents grow.

Technology3 sources

ByteDance Creates Dedicated AI Safety Team as Scrutiny of Autonomous AI Agents Intensifies

Technology3 sources

Prev12 3 4…44 Next