Latest news, developments and reporting on Large Language Models.
TechRadar published a step-by-step guide to self-hosting AI agents locally with OpenClaw and Open WebUI, reducing reliance on third-party cloud services.
Keenable raised $26M from Accel to build web search for AI agents; its 100B-document index is already in production.
Nearly 1 in 5 US LinkedIn users 'time travel' to add AI skills, study finds.
Perplexity launched Portable Computer, a local-first AI agent that runs models on-device, offering speed, security and cost savings. It's available now for paid Linux users with Nvidia RTX GPUs, coming to Windows in September.
A vulnerability in Nvidia's NemoClaw allows attackers to fully control a local Ollama server and poison AI agents via a single malicious website visit.
OpenAI-backed legal AI company Harvey launches Tenet, a model based on Chinese AI Kimi K3, highlighting US AI startups' growing reliance on Chinese open-weight models.
OpenAI CEO Sam Altman says AI has not yet had its iPhone moment and needs to learn Apple's focus lesson, while predicting a boom in small businesses that avoid AI.
Thomson Reuters launches Thomson-1, an in-house LLM based on Alibaba's Qwen, to cut costs and reliance on closed-source AI.
A new method distills directly from the original LLM, letting a 4-bit compressed model beat its full-precision checkpoint on 7 of 9 benchmarks.
Revolut has launched an AI research unit and an in-house foundation model built with Nvidia, reporting gains in fraud detection and credit risk accuracy.
An Amazon security engineer used Claude Opus to reverse engineer and modify firmware on multiple PC peripherals in 13 hours, warning the ease of AI-driven attacks poses a growing security risk.
Xiaomi released three self-developed chips including the 3nm O3 SoC, while Hugging Face explored a $13B sale and Alibaba launched Wan3.0 video generation model.
WRC 2026 panel exposes humanoid robot production bottlenecks; HiDream.ai launches world model; AFAC AI finals set for Shanghai.
A Pew Research analysis of 500,000 pages found AI authorship signs in 10% overall and over one-third of pages published after ChatGPT's release.
European VCs have named 19 startups working to lower AI's cost and energy demands as data-centre construction lags.
Alibaba released its video generation model Wan3.0 on August 24, capable of generating 30-second videos and supporting document inputs, with enterprise users praising its stability, realism, and texture.
OpenAI CEO Sam Altman said he fears AI will be controlled by a handful of companies or people, and warned against sacrificing liberty for safety. He also predicted a slower-than-expected boom in small businesses powered by AI.
Hugging Face is exploring a sale that could value the AI model hub at $13 billion or more, Business Insider reported.
Ox Alpha, a free AI coding model from an anonymous provider, has no identified builder, and code destination is unclear.
Wondershare Technology releases Filmora.TV, an AI-driven canvas workbench for TVC and brand video creation, partnering with tea brand Chayan Yuese for a demo video.
A mysterious AI model, Ox Alpha, released on OpenRouter, has sparked speculation about its maker, with theories ranging from Chinese labs to Microsoft.
A SiliconANGLE report says 88% of organizations suffered AI agent security incidents in the past year, urging stronger governance for autonomous systems.
Harvey released Tenet, its first post-trained legal AI, in research preview; it nearly doubles task completion on legal benchmarks.
An anonymous AI model, Ox Alpha, released on OpenRouter, has triggered widespread speculation about its creator, with evidence pointing to both Z.ai (GLM) and Google (Gemini).