Share
Linkedin
Twitter
Facebook
Whatsapp
Copy Link
OpenAI has signed four independent research groups into its internal safety testing. The access is deep, but the rules stay in OpenAI's hands.
Noeri raised $110 million led by Menlo Ventures, and its system tunes a company's AI setup continuously, with a person approving every change.
University of Pennsylvania engineers have built artificial muscle fibers that sense and correct their own movement, demonstrated on a piano-playing robot hand and a pair of robotic eyes.
Apple launched its Health AI Usage Study on Sunday, recruiting up to 200,000 Americans to test AI health answers built on peer-reviewed medicine. The move follows ChatGPT Health, Copilot Health and Perplexity Health.
A Brookhaven-led team says a self-learning AI can separate particle jets from background noise without labeled examples. It works in simulation so far; real data is next.
AI analysis of heart ultrasounds estimates a heart's age. A Mayo Clinic study presented at the ESC Digital & AI Summit links an older-than-real reading to higher death risk over 20 years.
Cornell's EchoMotion turns a smartwatch into a full-body motion tracker using ultrasonic echoes. No cameras, no extra sensors, and fitness features are next.
Uppsala University researchers developed GlioTrace, combining live brain tissue imaging with artificial intelligence to track thousands of individual glioblastoma cells. The method reveals how brain cancer cells switch movement patterns and respond to treatments, opening new pathways to understand tumor invasion.
OpenAI announced a major AI breakthrough on the Yang-Mills existence problem, one of mathematics' seven Millennium Prize Problems. A committee led by Nobel laureate David Gross verified the AI-generated proof using Lean formal verification, though the complete 1147-page proof remains unreleased and not fully understood.
Google DeepMind unveiled Gemini Robotics 2.5, its most advanced vision-language-action model for robotics, achieving a 96% success rate packing automotive parts at Mercedes-Benz. The AI model demonstrates human-level dexterity across 200+ tasks and rolls out through partners including Apptronik, Boston Dynamics, and Agile Robots.
Google DeepMind introduced AlphaGenome 2, an AI model that predicts how genetic mutations affect specific cell types by analyzing over 107 million individual cells. The model identifies nearly twice as many disease-associated variants as its predecessor, potentially transforming how researchers investigate conditions like inflammatory bowel disease.
Microsoft researchers revealed the Lazarus attack technique that transforms AI agents into sleeper backdoors by editing fewer than 10 dormant neurons in large language models. The modified models pass standard safety tests but execute attacker commands when triggered, highlighting critical vulnerabilities in AI security and the urgent need for model weight verification.
Researchers from Skoltech, Zelinsky Institute, and ITMO University published a comprehensive review in Chemical Society Reviews examining where AI in chemistry produces measurable value in research workflows. The study reveals that while AI methods in chemistry show promise in predicting properties and reactivity, planning chemical syntheses, and guiding automated experiments, high benchmark scores don't always translate to reliable laboratory performance.
New York Governor Kathy Hochul launched the nation's first state-run AI testing program, establishing the New York Center for AI Responsibility and Research at Binghamton University. Over 80 scientists will independently test AI systems for safety and security rather than relying on company self-regulation, with implementation tied to the RAISE Act's January 1, 2027 deadline.
Google DeepMind researchers used agentic AI to formally verify the Signal protocol's cryptographic guarantees in just four weeks. The breakthrough validates encryption protecting billions of messages on Signal, WhatsApp, and Google Messages while demonstrating AI's potential to transform software security verification.
Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Follow topics that matter to you and stay ahead.