AI Research news and analysis on Web Pulse: 9785 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
The US government lifted export controls on Anthropic's Claude Fable 5 and Mythos 5 models on June 30, after imposing them on June 12 due to a reported jailbreak. Fable 5 will be available globally starting July 1, while…
Panasonic Holdings' stock has more than doubled this year, giving it a record market valuation of ¥11.5 trillion ($71 billion), driven by rising demand for electronic components and batteries used in AI data centers. The…
Anthropic launched Claude Sonnet 5 as its new default mid-tier frontier model with a 1M-token context window and agentic capabilities, while also receiving approval to release Fable/Mythos 5 after government coordination…
Asia's manufacturing sector expanded in June driven by global AI demand, offsetting economic drag from the Iran war, with China, Japan, and South Korea reporting growth in factory activity due to robust demand for chips …
Google researchers have found that scaling laws similar to those for large language models apply to wearable sensor data, with validation loss scaling predictably with compute, model size, and data size. The scaling law …
Researchers introduced LLM4MOF, a closed-loop framework using language model agents to design metal-organic frameworks (MOFs) with high efficiency, achieving top-performing structures in just 400 property evaluations at …
Researchers introduced Multi-Block Diffusion Language Models (MBD-LMs) that enhance parallelism and training methods for faster, more accurate text generation. The MBD-LLaDA2-Mini model achieved a 78% increase in tokens …
New research spanning 11 models including Qwen 2.5, Gemma 2, and Llama 3.2 reveals that larger language models systematically shift their evaluation-awareness to earlier network layers, indicating they can recognize when…
Researchers introduced Symbolic Mechanistic Data Attribution (SMDA), a framework that links training data to high-level model behaviors, tested on Llama-3.2-3B-Instruct. SMDA revealed systematic safety gaps and biases, o…
NVIDIA is expanding its robotics hiring in China, opening roles in Beijing, Shanghai, and Shenzhen across embodied AI, simulation, deployment, and solution architecture. The company aims to develop next-generation genera…
A Siteline report testing a Claude agent on 100 B2B software products found that access errors and hidden pricing caused the agent to rely on third-party sites for information, with 30% of runs encountering errors and 5%…
Siteline found that AI agents frequently fail to extract list prices from B2B websites due to client-side JavaScript rendering and pricing pages behind contact forms, forcing agents to rely on stale third-party data. The…
A developer identifies 'context rot' as the gradual degradation of AI agent performance due to accumulated noise in the context window. The phenomenon causes recency bias, instruction dilution, stale state pollution, and…
Anthropic CEO Dario Amodei tempered expectations for AI-driven scientific breakthroughs at the company's 'The Briefing: AI for Science' event, suggesting the 'compressed 21st century' he predicted might not occur for a d…
A FIRST blog forecast reveals CVE volumes are tracking 46% above original projections, driven by AI-assisted discovery and structural reporting changes, with an estimated 66,000 vulnerabilities expected in 2026. The repo…
Microsoft released a report on AI diffusion across the United States, analyzing adoption trends and economic impacts. The study highlights varying levels of AI integration across industries and regions, emphasizing the n…
The San Francisco Board of Supervisors postponed a vote on a 362-page AI-assisted city code rewrite after concerns that the proposal was too broad and unclear. Supervisor Myrna Melgar used AI tool Claude to understand th…
Unit 42 researchers discovered that large language models consistently hallucinate web domains for legitimate brands, a phenomenon called phantom squatting. Adversaries are registering these nonexistent domains to interc…
Shanghai AI Laboratory's Agents-A1, a 35-billion-parameter Mixture-of-Experts model, matches or outperforms trillion-parameter rivals by training on longer task sequences instead of increasing parameter count. Released u…
Phoronix reported that Q2 2025 featured 872 news articles and 54 hardware reviews, with top stories including Linux kernel optimizations, NVIDIA Vera CPU, Intel Arc Pro B70, and AI/LLM impacts. Popular items covered Post…