Latest news, developments and reporting on Large Language Models.
At the 2026 China International Industry Fair, Hikvision showcased more than 100 industrial scenario digitalization products, from semiconductor inspection to fiber-optic belt monitoring, as it expands from security into industrial sensing and AI.
Mistral AI has released Voxtral 3, a family of three transcription models, one text-to-speech model and a LiveKit-based voice agent SDK. All four models are in public preview on the Mistral API.
A US magistrate judge has ordered Meta to identify the employees who approved or authorized downloading pirated books from Anna's Archive and Library Genesis over BitTorrent, rejecting the company's position that no one formally signed off, according to a TorrentFreak report.
Chery unveiled a 2035 strategy to become an AI-driven global technology company beyond carmaking.
DHS ordered federal agencies to stop using DeepSeek and remove its products from systems; China’s Commerce Ministry opposed the move.
Anthropic draft AI safety bill would let courts block models posing catastrophic risk, Politico reports.
DeepSeek open-sourced DeepInfer, its production inference framework for V4.1-Flash on Nvidia Blackwell and Huawei Ascend 950.
Inception AI says its Mercury 3 diffusion model beats GPT-6 Luna and Claude Haiku 5.5 on 8 of 12 tests at 4-8x speed.
Google is adding Personal Intelligence to Gemini in cars, drawing on Gmail, Calendar, Maps and Photos.
Hugging Face CEO Clem Delangue says Netflix’s ‘AI Gone Rogue’ misrepresents the OpenAI agent hack, TechCrunch reports.
OpenAI acquired Orchestra Bio and will add its team and technology to Rosalind, its life sciences AI model.
Astrolab’s FLIP lunar rover, slated for launch as early as November 2026, will carry HPE’s Spaceborne-4 computer to run custom large language models on the lunar surface, the first such use, MIT Technology Review reports.
NVIDIA released Cosmos3-H-Dreams, a real-time surgical simulator trained on clinical video and robot kinematics.
Electric Plant has raised $11.2 million in seed funding to commercialize sensors and Bombadil, an AI model trained to interpret electrical signals in about 125 plant species and relay their needs to growers, TechCrunch reported.
Hugging Face Blog has introduced FairStress, a 300-scenario benchmark that checks whether LLM decisions and their resistance to persuasive challenges vary by demographic identity across 29 domains.
Perplexity's Mac app can now run Computer fully on-device and offline, with a 36GB Apple silicon requirement.
VAST open-sources EdenLive, a real-time renderer that turns reference visuals and upstream game state into interactive video.
DeepSeek open-sources DeepInfer, the Apache 2.0 engine behind DeepSeek-V4.1-Flash, for Nvidia and Huawei hardware.
Eka Robotics, founded by an MIT professor and a former DeepMind researcher, has raised a $150 million Series A led by a16z and F-Prime at a $1.45 billion post-money valuation, betting that a force-driven model can make robot hands reliable enough for general use.
AI agents helped a hacker breach over 100 company sites in under a month; Deere and SharkNinja were among targets.
DHS orders federal civil agencies to remove DeepSeek products and bar future purchases, Chinese media report.
OpenAI says its models now handle COBOL and other legacy code better, citing migration tests for banks.
IBM-backed llm-d released version 1.0, adding multi-tenant controls, zero-downtime upgrades and agent-focused routing for self-hosted AI inference.
A Goodfire researcher's 26,000-word post argues LLMs may be easier to understand and control than thought, drawing both excitement and skepticism. Meanwhile, personal AI agents are prompting websites to block them, raising questions about the internet's future design.