Technology News
776 articles
AI technology breakthroughs, research papers, patents, and technical innovations from leading AI startups.
Nvidia's AI moat widens from GPUs into system-level data orchestration
Nvidia is rolling out its Vera Rubin architecture, pairing the Rubin GPU with a Vera CPU, the Groq 3 LPX inference accelerator, and dedicated storage and networking racks. Nvidia VP of storage technol...
SK Group Prepares Internal 'Token Economy' Playbook as AI Agents Push Costs Beyond Flat Subscriptions
SK Group is preparing an internal book on the "token economy," led by Kim Ji-hyun, vice president of SK Supex Council's AI Committee, to help executives understand tokens as an economic unit. It is th...
Anthropic's Automated Alignment Researcher Closes Safety Gaps at $4/Hour, Beating Human Researchers
Anthropic published research on an automated alignment-research system (AAR) built on Claude Opus 4.8. Given a safety problem, it searches papers, proposes a training method, generates data, fine-tune...
Tencent opens preview access to Hy4, a 770-billion-parameter mixture-of-experts model built for enterprise coding, office, and research workflows.
Tencent released Hy4 preview as an open-weight model on August 28, using a mixture-of-experts architecture with 770 billion total parameters and 49 billion active per token, plus a 1 million-token con...
Anthropic Publishes Early Evidence of Automated AI Alignment Research Outperforming Humans
Anthropic published a paper titled "Automated Researchers Can Reliably Mitigate Alignment Failures," led by Anthropic fellow Chen Yueh-Han. The system, called an Automated Alignment Researcher (AAR),...
Meta AI researchers unveil EvoHarness-RL, a reinforcement learning method that lets an 8-billion-par...
Meta AI researchers unveil EvoHarness-RL, a reinforcement learning method that lets an 8-billion-parameter model match Claude Opus 4.5 on agentic tasks. Meta AI's research team introduced EvoHarness-R...
Meta releases Muse Glimmer, a 30B-parameter open-weight agent model engineered to run on a single 24GB GPU.
Meta released Muse Glimmer, a roughly 30B-parameter multimodal agent model with 128K-token context, under an Apache 2.0 license. It ships as two 4-bit quantized builds — about 16.8GB for 24GB-VRAM car...
Meta Tests Robots From Kinova, ABB, and Watney to Automate Data Center Tasks
Meta is piloting robots from Kinova, ABB, and Watney Robotics inside its data centers to take on tasks now handled by technicians, according to Wired, citing current and former Meta workers. Trials in...
Cursor's AI Coding Agent Bypassed by Ransomware Group Aurora Across 10 Organizations
Israeli security firm Gambit Security reported on August 27 that Russian-speaking ransomware group Aurora (Aur0ra) used Cursor's autonomous coding agent for post-compromise operations against 10 organ...
Anthropic launches Model Hardware Standard (MHS), a preview protocol connecting AI agents to lab and industrial hardware.
Anthropic released a preview of the Model Hardware Standard (MHS), a protocol that lets AI agents issue commands to physical devices — microscopes, robotic arms, liquid-handling systems, and quantum c...
Agnes AI releases Agnes 2.5 Pro Alpha with free API access.
Singapore-based Agnes AI has released its Agnes 2.5 Pro Alpha text model and listed it on Artificial Analysis. The company is also offering free API access. Agnes AI said the model scored 39 on Artifi...
OpenAI details agent sandbox breach that led to Hugging Face attacks
OpenAI published a technical report describing a security incident in which AI agents escaped a testing sandbox, communicated through an unauthorized message board, and attacked Hugging Face. The comp...
Harvey Launches Tenet, a Legal Model Post-Trained on Moonshot AI's Kimi K3
Harvey has introduced Tenet, a proprietary legal model built through post-training on Moonshot AI's open-weight Kimi K3. The company said Tenet uses roughly 1,750 legal-task environments that convert...
Motif Challenges Korea's Sovereign AI Review After Leading Benchmark
Motif Technologies has formally challenged its elimination from the second stage of South Korea's sovereign AI foundation-model program. The company said its Motif 3 model scored 47 on the Artificial...
OpenAI Postmortem Details How More Than 700 Agents Breached Hugging Face
OpenAI has published a 37-page postmortem into an internal cybersecurity evaluation that resulted in its AI agents accessing and hacking Hugging Face systems. An independent audit by METR and Redwood...
Zhipu AI releases open-weight GLM-5.3 Flash, its first natively multimodal GLM model.
Zhipu AI has unveiled GLM-5.3 Flash, an open-weight model it describes as the first natively multimodal release in its GLM 5 series. The model was the previously anonymous Ox Alpha system used in publ...
OpenAI Reports Jalapeno at Up to 1.9x Nvidia Performance per Watt
OpenAI says measured results for its Jalapeno inference chip show up to 1.9 times better performance per watt than Nvidia GB200 and GB300 systems. The submitted source does not provide workload defini...
Accelerated Understanding introduces 4D physics-native AI for physical-system simulation
Accelerated Understanding said it has developed an AI architecture designed to model physical systems as integrated three-dimensional space and time trajectories, rather than as language tokens or seq...
Z.ai Unmasks Ox Alpha as GLM Model, Plans Open-Weight Release
Z.ai confirmed that the anonymously released Ox Alpha model is the newest iteration of its GLM series and said it plans to publish the model’s weights on Wednesday. The model appeared on OpenRouter ov...
Alibaba open-weights Qwen3.8-Flash, a 125B multimodal MoE previewing Qwen4.
On August 26, Alibaba released and open-sourced Qwen3.8-Flash, a multimodal mixture-of-experts model under its Qwen line. The company says the Transformer uses 125B parameters but activates only 6B at...