Research news and analysis on Web Pulse: 583 curated articles tracking the latest Research developments, tools, and research, updated continuously from vetted sources.
A lightweight (0.96 kg) portable robot was developed to provide isokinetic resistance training for juveniles with Spinal Muscular Atrophy (SMA) type II, integrating a variable stiffness mechanism for safe, customizable r…
Researchers developed a national inventory of 319,972 solar photovoltaic facilities and 91,609 wind turbines in China using satellite imagery and deep learning to assess solar-wind complementarity. The study found that p…
Dopamine is identified as a central regulator of long-lasting brain changes in mothers, driving persistent neural adaptations through a mechanism involving H3 dopaminylation. The study used transcriptomic profiling in mi…
Development of AI "scientists" designed to accelerate research by automating tasks like hypothesis generation and data analysis. It references recent studies showing that teams of AI agents can boost the speed of researc…
Large-scale neural activity in mice, except in the hippocampal area CA1, exhibits dynamics similar to those of a linear system governed by a critically normalized random symmetric matrix. This critical initialization pro…
AgentThreatBench is a new security benchmark that operationalizes the OWASP Top 10 for Agentic Applications (2026) into executable tasks, designed to measure whether an AI agent can be hijacked by malicious inputs in its…
Tool that removes both visible and invisible AI watermarks from images generated by models like Google Gemini, ChatGPT, and Midjourney. It strips metadata such as SynthID, C2PA credentials, and EXIF labels, and offers fe…
Anthropic's "Project Deal" experiment successfully demonstrated agent-to-agent commerce by having Claude AI agents autonomously complete 186 transactions worth over $4,000 in a closed, internal marketplace. However, the …
Following the Google I/O 2026 keynote, The Vergecast discussed the event's highlights, focusing on Google's new "agents" for AI utility, model updates including the Gemini Omni family, and the integration of more AI feat…
Hands-on introduction to the Futhark programming language through a series of commented example programs, organized by increasing complexity. It directs readers to the Futhark homepage and suggests alternative resources …
Mistral AI, a European AI leader, has acquired Emmi AI, an Austrian company specializing in Physics AI models for industrial engineering, in a strategic move to create a leading AI stack for sectors like energy, automoti…
According to a recent article, two new AI systems, Google's Co-Scientist and a system from FutureHouse, were introduced in Nature to assist researchers with tasks like drug retargeting. These "agentic" systems are design…
The article announces the release of OlmoEarth v1.1, a new family of Earth observation AI models that reduces compute costs by up to 3x while maintaining the performance of the original v1 model. This efficiency gain is …
Release of OlmoEarth v1.1, a new family of transformer-based AI models designed for processing satellite imagery, which reduces compute costs by up to 3x while maintaining the performance of the original v1 model. This e…
Google's SynthID, a digital watermarking technology for AI-generated content, has been adopted by major companies including OpenAI and Nvidia, expanding beyond its initial use by Google. SynthID has already labeled 100 b…
Google launched Gemini 3.5 Flash on Tuesday, a new AI model designed for autonomous agents that can independently execute coding pipelines and build entire operating systems, marking a strategic shift from conversational…
Gemini Omni is an AI creative studio that combines video editing with an understanding of physics, history, and narrative logic to generate coherent, realistic scenes from any input. It was developed with internal safety…
At Google I/O 2026, CEO Sundar Pichai announced that the company is entering the "agentic Gemini era," highlighting a surge in AI adoption with monthly token processing reaching over 3.2 quadrillion, a 7x increase from t…
At Google I/O 2026, the company introduced two new AI models, Gemini Omni and Gemini 3.5 Flash, with Gemini Omni capable of creating content from any input starting with video. The event also highlighted the advancement …
OpenAI announced two new measures to help verify AI-generated images: adopting the C2PA metadata standard and partnering with Google to integrate SynthID, an invisible watermark designed to resist tampering. The company …