AI News Archive - Browse Page 26 of 182
Browse AI news articles covering LLMs, tools, research, and industry trends
DOJ Cites National Security to Defend xAI Gas Turbines in NAACP Suit
The Justice Department wants a lawsuit dead. Its rationale hinges on two illegal gas generators in Memphis, Tennessee.
HPE AI Factory and NVIDIA unveil Vera, first CPU built for agents
HPE and NVIDIA just announced a new chip called Vera, which they are calling the first CPU built specifically for AI agents.
GLM-5.2 beats GPT-5.5 on SWE-bench Pro (62.1 vs 58.6) for 1/6 cost
For one-sixth the cost, GLM-5.2 just punched above its weight on SWE-bench Pro, scoring 62.1 to GPT-5.5’s 58.6.
3D UMAP of 50k Validation Embeddings Shows Industry and Zip Code Clusters
Fifty thousand transaction embeddings, compressed into a three-dimensional space, reveal something striking: merchants cluster by industry, users by...
PrologMCP Launches as Task-Agnostic Open-Source Server for LLM Agents
Any developer who’s wired modern AI into Prolog knows the drill. It’s a custom job every single time. You build the translator.
Berlin court says Google’s AI Overviews are search format, not content
Germany is locked in a characteristically meticulous legal debate. The core question: who answers when a robot invents facts?
SpaceX to buy Cursor for USD 60 billion as platform struggles to break even
Sixty billion dollars for a company that doesn't make money. Only in AI. SpaceX is buying Cursor, the coding tool, for that price.
Reconfigure OpenClaw on Mac Mini to Deploy a Local LLM Model
Your Mac Mini is a powerhouse. But it’s not enough to have a local LLM running, you need OpenClaw to see it, talk to it, and route your requests.
Roadmap to LLM Engineer in 2026: Foundations, Prompting, Fine‑Tuning, Alignment
Forget the hype. The job of an LLM engineer is mostly grunt work. You're not teaching a model to be clever.
Attention output GEMM reduces blended Fprop speedup to 1.47× in NVFP4 training
The 1.47x speedup for NVFP4 training isn't fake. It's just the real answer after you pay the bill.
NVIDIA ACE SDK Enables On‑Device AI Companions via UE5 Plugins
The cloud is dead for NPCs. NVIDIA is moving the AI brain from distant servers directly into your graphics card.
AMD builds Llama 3.1 8B pretraining benchmark for MLPerf, using random weights
AMD has submitted a training benchmark for a model that doesn't learn. For the latest MLPerf results, the company pretrained Meta's Llama 3.1 8B...
AMD's MI355X CDNA4 GPU Shows Competitive Training Times in MLPerf v6.0
AMD just matched Nvidia’s top chip. In a head-to-head sprint, the company's new MI355X accelerator fine-tuned a Llama 2 70B model in the same time as...
NVIDIA Blackwell Leads MLPerf Training 6.0 with Full‑Stack Scale
Nvidia won everything. The MLPerf Training 6.0 benchmark results are in, and the company's Blackwell platform took first place in every single test.
Estonian institute benchmarks AI models' vulnerability to Russian propaganda
The Institute of the Estonian Language just graded sixty AI models on a critical new subject: Kremlin propaganda.
100+ security experts warn Fable 5 export ban handcuffs defenders, not attackers
One hundred security experts just signed a letter that cuts through the noise: the government’s Fable 5 export ban doesn’t stop attackers, it only...
Study quantifies AI agent trust formation, breakage, recovery in survival game
Every AI company wants you to think its models are responsible team players. A new study suggests the biggest models are, sort of, but not in the way...
Malaysia’s Respond.io raises USD 62.5M to expand AI messaging, target acquisitions
Another messaging startup just raised a mountain of cash. This one might actually know what it's doing.
QPILOTS Offers Test‑Time Q‑Steering for Flow Policies, Avoiding Gradient Loss
Building a flawless AI model is tough. Getting a reliable answer from it can be even tougher.
DR-DCI Enables Agent-Callable Retrieval to Expand Local Workspace Efficiently
Most AI research papers sell you a new way to fail. They’ll claim they’ve cracked some fundamental tension, then quietly fudge the data.