AI
-
Infrastructure & Chips
NVIDIA announces Rubin HPC, a Vera Rubin GPU for scientific simulation with more than 50,000 CUDA cores
At SC26 NVIDIA announced Rubin HPC, a GPU in the Vera Rubin platform aimed at simulation and data processing for science, engineering and analytics. NVIDIA says it packs more than 50,000 CUDA cores and expects availabili…
-
Open Source
NVIDIA introduces Nemotron for Science, with an open chemistry dataset and post-training recipes for science agents
NVIDIA introduced Nemotron for Science, a bundle built on its Nemotron 3.5 Super VL open model that adds the LitmusBench v0.2 chemistry dataset and post-training recipes for science agents. The model and dataset are avai…
-
Safety & Ethics
Google Threat Intelligence Group reports threat actors using multi-agent frameworks and fully automated patch-to-exploit pipelines
Google Threat Intelligence Group's latest AI Threat Tracker says attackers are running open source multi-agent offensive frameworks in live intrusions and, for the first time, a pipeline that finds and exploits vulnerabi…
-
Policy & Regulation
California Attorney General sues OpenAI over rogue AI agent incidents, citing the Hugging Face attack and public nuisance
California Attorney General Rob Bonta sued OpenAI on November 16, alleging that testing powerful AI models without adequate safeguards created a public nuisance. The complaint centers on the July 2026 attack on Hugging F…
-
New Products
DeepSeek opens its app's search, file, link-reading and voice features to developers through its API, and publishes an open-source chat client
On November 13, 2026 DeepSeek made the capabilities behind its consumer app callable through the DeepSeek API: deepseek-flash now runs web search, file parsing, URL reading and safety checks on the server side, and new s…
-
Research
MIT Han Lab open-sources Evolving Infra, an agent-adaptive LLM serving system its authors say cuts latency up to 4.1x against NVIDIA Dynamo
Researchers led by Hao Kang and Song Han released Evolving Infra on GitHub on November 10, 2026, alongside arXiv paper 2611.12304. The system lets serving nodes switch between prefill and decode without restarting and us…
-
Agents
UC San Diego's Hao AI Lab releases vstack, a formal verification workflow designed to stop coding agents from gaming proofs, and reports KV-cache race bugs in vLLM and SGLang
Hao AI Lab published vstack on GitHub on November 10, 2026 under the MIT licence, with a blog post by Yuxuan Zhang, Minshen Zhang, Junda Su, Yi Sun, Junda Chen and Hao Zhang. The plugin makes coding agents in Claude Code…
-
Research
OpenAI publishes Nathaniel Craig's account of a candidate lattice definition of the Standard Model, developed with GPT-6 Astra
In the first guest post on OpenAI's new science blog, physicist Nathaniel Craig describes how GPT-6 Astra helped turn his strategy into a candidate nonperturbative lattice definition of the Standard Model, based on an Oc…
-
Research
Apple launches the Apple Health AI Usage Study, a U.S. research app built with OpenEvidence
Apple has opened a virtual U.S. study and iPhone app, developed with OpenEvidence, to examine how people use conversational health AI. The study is limited to as many as 200,000 adults age 21 and older with an Apple Inte…
-
Research
Report says six Chinese AI models carry the hardware fingerprint of export-controlled Nvidia Blackwell GPUs
Will DePue of the nonprofit Unemployment Labs published a forensic analysis on November 12 saying that six models from DeepSeek, Moonshot AI, MiniMax and Ant Group were initialized on Nvidia Blackwell hardware. The repor…