#ai Tech News.

Mogu IoT, a Guangzhou-based industrial AI company, closed its Series C C1 and C2 rounds, raising nearly 200 million yuan. Yunhui Capital led the round, with Yuanhe Origin participating. The funds will develop the Lingzhi AI model and physical AI agents for energy efficiency.

Catch emerged from stealth with $5 million in seed funding co-led by Entrée Capital and Pitango, with Seedcamp and Factorial Capital participating. Founded by CEO Nir Sabato and CTO Yoav Ramon, the startup builds autonomous AI agents for executive administration, integrating with Gmail, Slack, and Outlook.

OpenAI is spending $1 billion to subsidize access to its Daybreak security platform after unveiling Astra, an AI that can independently find previously unknown software flaws and turn them into working attacks.

OpenAI launched GPT-6 Astra, a new frontier AI model for computer use, professional work, science, coding, and cybersecurity. The model can browse the web, build websites, and handle complex tasks with less human intervention, while presenting significant cybersecurity capabilities and risks.

Hugging Face, founded in 2016 as an emoji-named teen chatbot app, became an open-source AI advocate. CEO Clément Delangue approached Nvidia this summer to pursue a deal.

Andreessen-backed AI startup Gimlet is now valued at $3 billion. The valuation marks a significant milestone for the company, which has attracted major venture capital backing.

Soitec shares jumped after the company upgraded its guidance, citing strong AI demand. The guidance upgrade reflects increased expectations driven by artificial intelligence market growth.

Vals AI analysis finds open-weight models performing multi-stage tasks, like building a web app, can have an environmental impact 10,000 times greater than simple queries. Multi-stage tasks require exponentially more energy or water than simple queries do.

Nvidia agreed to acquire Hugging Face for $12.93 billion. Hugging Face hosts over 3 million AI models and serves more than 18 million developers and 200,000 companies. Nvidia says the platform will remain open.

Nvidia launched PAIR, a free open-source tool that pools idle computers on a home network for local AI inference and agentic workloads. It supports GeForce RTX 20-series and newer, RTX Pro GPUs, DGX Spark systems, and Apple M4 chips or newer, with beta available for Windows, Linux, and macOS.

Nvidia agreed to acquire Hugging Face for $12.9 billion, with CEO Jensen Huang promising the platform keeps its brand, leadership, and neutrality. Hugging Face CEO Clem Delangue aims to grow users from about 18 million to over 100 million.

OpenAI pledged $1 billion in subsidized access, training, and technical support for cyber defenders through its Daybreak for Frontline Defenders initiative. The program initially targets U.S. operators of essential services, including water utilities, electric grid operators, and community banks, with plans to expand to partner countries.

OpenAI released GPT-6 Astra, its most intelligent model, initially to Trusted Access Program members, with Plus, Pro, Business, and Enterprise subscribers gaining access in coming days. It excels at computer-use, browsing, software engineering, and cybersecurity, scoring 98.6% on ARC-AGI-3 and a perfect ExploitBench score.

Madrona research shows 74% of 150 enterprise IT professionals plan to expand AI budgets, yet fewer than half of AI pilots reach production. 77% re-evaluate AI vendors every six months, making startup ARR less secure than traditional SaaS contracts.

OpenAI commits $1 billion in subsidized model access, training, technical support, and partnerships to a new initiative protecting essential services worldwide. The funding expands subsidized access to models for participating organizations.

Grok 4.6 from xAI outworks ChatGPT and Claude in daily use, offering built-in image and video generation, concise responses, and efficient coding in Cursor. It lacks Claude's Artifacts and native desktop apps, but Grok 4.7 is upcoming.

Sen. Bernie Sanders announced legislation for a permanent ban on superintelligence and a pause on advanced AI development. The bill, co-announced with Rep. Greg Casar, would create a federal regulator and impose up to 20 years in prison for violators. OpenAI released GPT-6 Astra the same day.

Meta stopped grading employees on AI tool usage, replacing it with impact-based reviews. Workers now test the agentic AI tool Hatch, which browses the web and operates apps. Some employees avoid connecting personal accounts due to privacy worries, while token consumption surges.

Nvidia's Personal AI Router (PAIR), announced at IFA 2026, is a free, open-source beta tool that distributes AI workloads across idle local PCs running Windows, macOS, or Linux, preventing GPU competition and performance slowdowns. It supports Nvidia GeForce RTX 20 Series and newer, RTX PRO workstation GPUs, DGX Spark, and Apple M4 or newer.

Microsoft AI launched MAI-Transcribe-2, a speech recognition model that outperforms Gemini 3.5 Transcribe and GPT-Transcribe. The model is priced at $0.10 per audio hour through 2026, following early-bird price cuts of roughly 72%.

ChatGPT, Claude, Gemini, and Grok suffered outages starting around 9:30 a.m. ET on Sept. 3. Anthropic attributed its disruption to infrastructure issues, restoring service by 12:16 p.m. ET. Microsoft Azure problems coincided with the incidents, though no common cause was confirmed.

Meta Superintelligence Labs released Muse Spark 1.3, its fourth Muse Spark release in five months, targeting long-horizon agentic and coding work. It ships in Muse Code and the Meta Model API, but weights remain closed and max reasoning is safety-gated.

Protein Industries Canada is investing up to $2.2 million in two AI AgTech projects. One project pairs TerraVision360, Metos Canada, and Rocky Mountain Equipment to predict Ascochyta blight risk. Another partners SuperGeoAi Technology with Southview Farms on a LiDAR drone for measuring grain volumes.

Qwen 3.8 27B is now available on Cerebras, delivering inference at 1500 tokens per second. The deployment brings the open-weight model to Cerebras hardware for high-speed processing.

OpenAI, Anthropic, xAI, and Google AI models experienced overlapping service interruptions Thursday morning. Anthropic reported a partial outage with elevated errors on Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5 at 9:23 am, deployed a fix, and resolved the issue by 12:16 pm.

ChatGPT, Claude, and Grok all suffered service disruptions on September 3, with Microsoft Azure reporting a network infrastructure incident in its East US region during the same period. OpenAI, Anthropic, and xAI confirmed issues and began recovery efforts, though no common root cause was confirmed.

Google launched Gmail Live, Docs Live, and Keep Live, Gemini-powered conversational voice features for searching inboxes, drafting documents, and organizing notes. Gmail Live serves Google AI Plus, Pro, and Ultra customers; Docs and Keep Live serve Pro and Ultra users. All features support English on iOS and Android.

Google DeepMind and Google Research released WeatherNext 3, an AI weather model with 5-kilometer resolution, hourly forecasts, and up to 60% improved precipitation accuracy. It rolls out to Search, Maps, and Gemini, and outperforms models from Microsoft, Nvidia, and ECMWF on Operational WeatherBench.

OpenAI, Anthropic, Google, Microsoft, Amazon, Oracle, and over 100 other companies signed a letter urging a global surge in cyber defense, warning that AI-enabled attacks will become more widespread and sophisticated in coming months, putting hospitals, water plants, and internet infrastructure at risk.