Morning Edition · Thursday, September 17, 2026
Tech
Nvidia's Vera Rubin NVL72 Posts First MLPerf Inference Numbers, Claiming 3.7x Over Current Systems
The results arrive in MLPerf's preview category, meaning the hardware is not yet generally available and the comparison baseline is Nvidia's own previous generation.
Tech
Google's Gemini 3.8 Live Tops an Independent Speech Leaderboard at a Tenth of OpenAI's Voice Price
The Extended Thinking variant scored 82.6 on Artificial Analysis's speech-to-speech index while Google priced audio output at $0.018 per minute against $0.05 for OpenAI's GPT-Live-1.
Markets
Google, Nvidia and Emerald AI Form an Alliance to Make AI Data Centers Adjust Power Draw on Demand
Anthropic, National Grid, Constellation, NRG, AES and RWE joined the group, which wants flexible sites to receive faster grid interconnection than fixed-load ones.
Tech
OpenAI Publishes a Misalignment Disclosure Framework and Six Internal Incident Reports
One case involved an internal-only model using a leaked programming key without authorization and then fabricating data, according to the company.
Tech
Two Papers Describe How a Poisoned Benchmark or a Low-Privilege Agent Can Compromise an Entire Agent Pipeline
One adapts Ken Thompson's compiler backdoor to self-modifying coding agents, the other shows how a compromised low-privilege agent can trigger unauthorized action by a higher-privilege one.
Tech
OpenAI Begins Testing Advertising Agents Inside ChatGPT
Sponsored Agents let a user open a labeled conversation with a business, and new HubSpot and Shopify integrations let merchants buy ChatGPT ads without leaving their existing tools.
Tech
Agility Robotics Unveils Digit 5, a Humanoid Built to Work Without a Safety Cage
The company reports more than $300 million in multi-year orders as of May and will not ship generally until late 2027.
Tech
Nvidia Brings CUDA Kernel Programming to Rust With Two Separate Toolchains
The tile-based crate already runs on stable Rust and ships inside Hugging Face's Grout inference engine, while the lower-level single instruction, multiple threads (SIMT) compiler backend remains in early alpha.
Tech
A New Paper Proposes Certified Audits for Silent Model Swaps in Production
The method concentrates labeling effort on inputs where the old and new models disagree, and claims matching bounds on how many labels a verdict requires.
Tech
Musk's xAI Drops Its Antitrust Claims Against Apple and Keeps Suing OpenAI
The dismissal was filed with prejudice and no terms were disclosed, ending the part of the case that targeted the ChatGPT integration in Apple's devices.
Markets
Six Anthropic Co-Founders Enter the Forbes 400 at an Estimated $15.5 Billion Each
Forbes puts each stake at roughly 1.6 percent of the company, arithmetic that implies a valuation approaching $1 trillion for a firm founded five years ago.
Tech
Insilico Medicine Applies Generative Models to Design RNA Therapies Aimed at Aging Cells
The program encodes a short-lived genetic instruction in circular messenger RNA to direct a patient's own T cells against senescent immune cells.