AISep 21, 2026NVIDIA Blog
Why Deploying Physical AI at Scale Demands Safety at Every Layer
NVIDIA is pitching Halos as full-stack safety for cars and robots — because once AI moves steel and wheels, a model card is not enough.
Desk
AI
Broad artificial intelligence coverage.
AISep 21, 2026NVIDIA Blog
From Enablement to Execution, Egypt’s AI Ecosystem Reaches Production Scale
Egypt’s AI story is shifting from potential slides to onshore AI factories — and NVIDIA is writing itself into the ribbon-cutting.
AISep 21, 2026Ars Technica
Trump rejects AI slowdown calls, launches "AI Force" instead
While lab CEOs circle a rare slowdown consensus, Trump is naming an AI tsar, floating an “AI Force,” and calling safety alarms a hoax.
AISep 21, 2026Microsoft Research
Improving synthesis prediction of small molecules at scale with RetroChimera
Microsoft Research’s Nature paper on RetroChimera pairs complementary retrosynthesis models and open-sources the stack—useful chemistry acceleration that still needs lab proof beyond chemist preference tests.
AISep 21, 2026NVIDIA Developer
Turn Your Latest Observations Into Timely Weather Decisions With NVIDIA Earth-2
NVIDIA’s Earth-2 tutorial shows how to fold live observations into AI weather models without full retrains—vendor metrics look strong, and operators should demand matched proofs on their own assets.
AISep 21, 2026NVIDIA Blog
AI Security Is an Engineering Problem — How to Solve It at Every Layer of the Agent Stack
NVIDIA’s Saša Zdjelar frames agent security as requirements, controls, owners, and evidence—not vibes—and the OpenShell runtime push is the part that can actually fail closed.
AISep 21, 2026TechCrunch
Google’s $899 Googlebook is a bet that you’ll buy a new laptop for Gemini
At $899 with Magic Cursor and a year of AI Pro, Googlebook is less a Chromebook refresh than a hardware funnel into Gemini—especially the K-12 fleet that already lives on Chrome.
AISep 21, 2026MIT Technology Review
4 ways to address the failures we found along the US border’s “virtual wall”
MIT Technology Review’s investigation found people dying near AI-enabled border towers that missed them or that agents ignored—and the fix starts with treating those deaths as system failures, not bad luck.
AISep 21, 2026OpenAI
Advisory Group on Mathematics and Artificial Intelligence
OpenAI says an internal model resolved Navier–Stokes and 100-plus open problems — and is standing up an independent math advisory group to help tell the field.
AISep 21, 2026OpenAI
Higgsfield AI ships new video features in a day with GPT-6 Astra
Higgsfield says GPT-6 Astra lets one engineer ship new video-ad exploration features in a day — a customer story about agent speed, not a new model launch.
AISep 21, 2026The Verge
UN says AI safeguards can’t wait for certainty
The UN’s first major scientific brief on the Hugging Face agent hack argues that loss-of-control risk is exactly when the precautionary principle applies—act before the mechanism is fully mapped.
AISep 21, 2026NVIDIA Blog
5 Companies Using NVIDIA AI for Clean Energy
At Climate Week, NVIDIA is selling a clean-grid pitch that also solves its own power problem — and a few of the numbers are hard to ignore.