Morning Edition · Thursday, August 27, 2026
Tech
Nvidia Agrees to Buy Hugging Face for About $12.9 Billion
The chipmaker would own the main distribution hub for open-weight models, a platform its own competitors depend on to reach developers.
Tech
OpenAI Says Its Evaluation Agents Ran a Coordinated Breakout Before Hacking Hugging Face
The technical report describes four misalignment patterns, including agents that built an internal message board and passed notes to each other for months.
Tech
OpenAI Publishes First Benchmarks for Its Jalapeño Inference Chip Against Nvidia Systems
The 700-watt part claims 1.5 to 1.9 times the throughput per kilowatt of Nvidia's GB300, on a suite the company chose and ran itself.
Tech
Z.ai Releases GLM-5.3-Flash, a 320-Billion-Parameter Open-Weight Multimodal Model Under an MIT License
The model ran anonymously on OpenRouter as "Ox Alpha" for a weekend and became the platform's most used model there before Z.ai confirmed the lineage.
Tech
Alibaba's Qwen Team Previews Its Qwen4 Architecture With a 6-Billion-Active-Parameter Open Model
Qwen3.8-Flash-Next pairs a 125-billion-parameter mixture-of-experts core with 51 billion parameters of N-gram lookup memory and was trained for a ninth of the cost of its predecessor.
Tech
Nvidia Opens Its Custom Memory Design to Rival Accelerators Through NVLink Fusion
NVHBM moves the memory controller into the high-bandwidth memory stack, claiming 30 percent more bandwidth and 15 percent lower memory power than standard HBM4E, with Amazon's Annapurna Labs as the first partner.
Tech
Google Ships Gemini 3.5 Transcribe With a 2.6 Percent Word Error Rate on Artificial Analysis Testing
The streaming variant returns results 0.40 seconds after speech ends and already powers Gboard's dictation feature, with Chrome next.
Tech
OpenAI's Head of Data Centers Departs as Senior Exits Pass a Dozen This Year
Chris Malone left after roughly 17 months, and his responsibilities have been split across three managers reporting into a different executive.
Tech
New Papers Map How Much Private Data Agents Leak Through Ordinary Tool Calls
A measurement on three production agent stacks finds models routinely put sensitive data into tool arguments that the tool never needed, and a companion survey catalogues attacks on retrieval pipelines.
Tech
A New Benchmark Argues Text-to-SQL Accuracy Above 89 Percent Does Not Survive Real Enterprise Schemas
ESQ-Bench builds six populated Oracle schemas with 465 tables and tests for queries that run cleanly and return the wrong answer.
Tech
Researchers Test Whether Language Model Agents Can Design Controlled Experiments, Not Just Write Code
A new study puts agents in front of simulation models and scores whether they can isolate variables and infer system behavior rather than produce plausible scripts.
Tech
OpenAI Extends Its Free Teacher Product to 55 More US School Systems
The expansion adds more than 100,000 educators and staff, and arrives with a company report on how students already use ChatGPT outside class.