AI News Daily 03-10

AI News | Daily Briefing | Aggregated Data | Frontier Science | Industry Voice | Open Source Power | AI & Human Future | Access Web Version ➡️ | Join the Chat! 👋

Today’s Digest 🔥

Microsoft Phi-4 multimodal, Tencent launches Penguin-VL vision-language model
Drosophila whole brain with 125,000 neurons digitally uploaded, Claude solves a thirty-year problem in one hour
NVIDIA reports large-scale enterprise AI deployment, Gartner predicts spending to hit $2.52 trillion
Alibaba page-agent, NousResearch Hermes Agent and other open-source projects go live
Oracle lays off 30,000 to pivot to computing power, ByteDance Doubao AI phone and Tencent WorkBuddy make their debut

Product & Feature Updates

1. Microsoft releases Phi-4 multimodal reasoning model. Microsoft’s brand-new Phi-4 inference model has officially dropped! This bad boy boasts powerful multimodal reasoning capabilities (AI News) and its upgraded inference scaling technology is seriously impressive. ✨ Expect significant boosts in healthcare and operations & maintenance scenarios, and guess what? Video generation capabilities are now live! 🎉

2. Claude Code launches loop automation commands. Anthropic just dropped a huge update for Claude Code! It now features brand-new, built-in loop commands for task automation (AI News) , letting developers efficiently orchestrate 🛠️ repetitive tasks. And hey, xAI also launched its Speech Synthesis API around the same time, supporting the generation of diverse artificial voices! 🗣️

3. Cursor announces pivot to self-developed models. Cursor is officially shifting gears, announcing its pivot to a proprietary model development path! They’re teaming up big time with Meta and NVIDIA (AI News) to supercharge iteration, and as a result, code generation quality has seen a massive optimization 🔧. This move marks a fresh evolution in the AI programming tool arena, with Cursor’s self-developed route looking to open up a significant competitive gap! 💪

4. Honor releases MagicAgent intelligent agent foundational model. Honor just dropped some big news with its MagicAgent foundational model! This full-scenario generalized planning agent model (AI News) boasts 30 billion parameters, surprisingly rivalling models with hundreds of billions! 🔥 It supports heterogeneous orchestration and automated task execution. Meanwhile, Alibaba also launched its QoderWork desktop agent, which is now fully covering both Mac and Windows platforms! 🎉

5. Tencent launches Penguin-VL vision-language model. Tencent has officially unleashed its Penguin-VL vision-language model, built on a new vision-language processing architecture (AI News) ! This model’s image understanding capabilities are incredibly robust 📸, making visual and text cross-modal fusion super smooth. It’s ready to power more domestic applications in various scenarios, and the technical breakthrough in visual reasoning is seriously impressive! 😎

Frontier Research

1. HKUST AudioX unified audio generation model. HKUST has cooked up something pretty cool: the AudioX unified audio generation model! This unified audio generation framework AudioX (AI News) uses just one model to handle all tasks, including music scoring and more! 🎶 Trained on a whopping 7 million samples, it’s already open source and hit SOTA levels on multiple benchmarks, earning it a spot at ICLR'26. Want the nitty-gritty? Check out the full paper (AI News) .
   
   

2. First cyber Drosophila whole brain successfully uploaded and revived. Hold up! The first-ever cyber Drosophila whole brain has been successfully uploaded and revived! Scientists achieved the incredible feat of Drosophila whole-brain neuron digitalization upload (AI News) , with 125,000 neurons now powering a digital body to move autonomously! 🧠 This isn’t just traditional algorithms; it’s genuine whole-brain simulation technology. Even Elon Musk was amazed, tweeting “Wow!” (AI News) . This breakthrough genuinely ushers in a new, mind-blowing era for brain science and AI integration! 🤯
   
   

3. Claude cracks Knuth’s thirty-year problem in one hour. Talk about brainpower! Claude just cracked Donald Knuth’s thirty-year problem in just one hour! The Turing Award winner himself was deeply shocked by this breakthrough (AI News) . Claude 4.6 managed to solve the thirty-year mathematical mystery in just sixty minutes ⏱️ by deriving a general algorithm using a ‘snake-like construction.’ Knuth literally tipped his hat to AI, exclaiming ‘Shock!’ 🤯 This is a massive leap, pushing human-AI collaboration in problem-solving into a new era (AI News) .

4. M4 chip’s hidden computing power cracked by Claude. Big news! Claude has successfully cracked the Apple M4 chip ANE hidden computing power (AI News) ! Get this: the M4 chip’s ANE energy efficiency is a mind-blowing 50 times that of the H100! 🤯 This means the Mac mini could soon be training large models with super low power consumption. Human-AI collaboration is kicking off a whole new paradigm for on-device training, shaking off software limits and letting private AI models evolve right on your desktop! 🚀

Industry Outlook & Social Impact

1. NVIDIA releases 2026 AI State Report. Good news from the green team! NVIDIA just dropped its annual AI industry status report (AI News) , revealing that businesses are ditching pilot programs and jumping straight into large-scale AI deployment. Apparently, AI has officially become the go-to infrastructure for revenue growth! 📈 Companies are seeing massive cost cuts and huge leaps in productivity. Basically, AI application penetration across all industries is soaring! 🔥

2. a16z releases March Top 100 AI Consumer Applications list. Heads up! a16z just dropped its latest Top 100 Generative AI Consumer Applications list (AI News) for March! Turns out, Sora’s downloads are seeing stable growth 📈, while Genspark made a strong entry after snagging $300 million in funding. The global AI consumer market is now splitting into three main camps, so expect fierce shake-ups in consumer-grade application rankings! 💥

3. Netflix fully acquires Affleck’s AI film studio. Plot twist! Netflix just fully acquired Ben Affleck’s AI film production company (AI News) , and streaming giant Netflix wasted no time! 🎬 This move signals a huge transformation for the film industry towards AI content generation, with production costs likely to plummet! 💰 Hollywood’s dive into AI is suddenly speeding up big time.

4. Gartner predicts global AI spending to reach $2.52 trillion. Buckle up, because Gartner predicts global AI spending will hit $2.52 trillion! Their latest forecast report (AI News) shows a whopping 40% increase in spending, with this massive $2.52 trillion market rapidly taking shape! 💸 We’re already seeing Agentic AI autonomously completing tasks, but here’s the kicker: nine out of ten pilot projects are still not profitable. Looks like the market is slowly but surely shifting from frenzy back to rational judgment. 🤔

5. White House predicts AI will surpass Industrial Revolution’s impact. Woah, big claims from the White House! They’re predicting AI’s growth will surpass the impact of the Industrial Revolution! According to their latest policy interpretation (AI News) , GDP could even double! 💰 Meanwhile, the UK is pouring serious cash into building sovereign computing power infrastructure, and banks are eagerly launching AI centers. Looks like automated processes are expanding full throttle across all sectors! 💪

Top Open-Source Projects

1. Alibaba open-sources page-agent web intelligent agent. Check it out! Alibaba just open-sourced its page-agent web intelligent agent! This newly launched web control agent tool (AI News) lets users directly control web pages using natural language! 🌐 It’s super lightweight, implemented purely in JavaScript, and specifically designed to tackle the pain points of automating complex interactions. Developers, head over to GitHub for full source code (AI News) to give it a spin!

2. NotebookLM Python interface library launches. A new NotebookLM Python interface library just dropped, and it’s incredibly powerful! 🐍 This programming access tool library (AI News) allows Python to fully control all NotebookLM features, successfully breaking through those official web-based usage limitations. It also supports connecting various AI Agents for sweet automation. Plus, developers can even use CLI commands to handle full-process operations (AI News) !

3. NousResearch open-sources Hermes Agent. Big news from NousResearch! They’ve officially open-sourced their Hermes Agent! This evolvable intelligent assistant (AI News) is designed to continuously evolve 📈 alongside your user habits. The key highlight is its complete logic loop design, aiming to create a truly growable personal assistant! 🌱 Ready to try it? Grab the code from the project address (AI News) !

4. CodeGraphContext code graph MCP launches. Say hello to CodeGraphContext! This open-source project, boasting ⭐1.5k stars, turns your code into a graph, ensuring AI Agents no longer have to blindly guess code logic relationships! 🗺️ There’s even a web playground (AI News) live for you to test it out. It connects directly to GitHub repos or local projects, making code symbols and dependencies crystal clear! ✨

Social Media Shares

1. OpenAI officially announces acquisition of Promptfoo. Big news from OpenAI! They’ve officially announced the acquisition of Promptfoo (AI News) , an AI security detection platform. This strategic move aims to supercharge OpenAI’s AI security defense system! 🔒 Promptfoo can automatically fix system vulnerabilities during development and help companies pinpoint hidden security risks. This is a big deal for the entire industry! 🚀

2. ByteDance Doubao AI phone’s MWC global debut. ByteDance’s Doubao AI phone made its global debut at MWC! Teaming up with ZTE, ByteDance showcased the Doubao AI phone preview version (AI News) . This phone boasts system-level permissions, allowing it to tap across apps! 📱 However, Tencent’s Pony Ma publicly voiced concerns about its security, sparking a heated debate. The big question is: where do we draw the line between privacy boundaries and convenience? 🤔 This underlying ecosystem disruptor (AI News) has definitely ventured into uncharted territory.

3. Tencent releases WorkBuddy full-scenario intelligent agent. Get ready for a smarter workday! Tencent has officially released WorkBuddy, its full-scenario office intelligent assistant (AI News) ! It’s deeply compatible with the OpenClaw protocol 🔌, making it super easy to use right out of the box. Users can even one-click switch between various mainstream domestic large models (AI News) and connect to multi-client collaboration platforms like WeCom and Feishu. Basically, PC office work is officially entering a full-blown autonomous driving mode! 🚀

4. Oracle lays off 30,000, pivots to computing power. Whoa, big news from Oracle! The tech giant suddenly announced a layoff of 30,000 people (AI News) , sending shockwaves through the industry. But hold on, it’s not because AI has replaced human jobs 🙅‍♀️. Instead, their high salary budget was entirely reallocated – a massive chunk of cash went straight into buying NVIDIA chips! It looks like in this insane computing power race, ’layoffs for compute’ is sadly becoming the norm. 🤷‍♀️

5. ByteDance VolcEngine launches cloud-based ArkClaw. ByteDance’s VolcEngine just launched its cloud-based ArkClaw! This cloud-based all-in-one ArkClaw service (AI News) is designed as an out-of-the-box cloud SaaS model ☁️. It’s perfectly adapted for Feishu collaborative office, offering lightning-fast responses, and has integrated mainstream large model libraries to double efficiency. Plus, you can quickly experience it (AI News) for just 9.9 RMB for the first month!

6. OpenClaw coding agent practical ranking released. Alright, tech enthusiasts, the OpenClaw Coding Agent Practical Ranking is out, and it’s sparking some serious chatter! 🏆 Gemini 3 Flash snagged the top spot for coding practical ability (AI News) , with the Claude family performing super stably and dominating the charts. But here’s the shocker: GPT-5.2 unexpectedly fell behind, landing with the lowest success rate! 📉 The real-world gaps between these models are pretty surprising!

7. NVIDIA releases new AI inference processor. Get ready for a speed boost! NVIDIA just unveiled its new generation of AI inference processors! This AI inference specific processor (AI News) integrates Groq chip technology ⚙️, sending performance skyrocketing. Even OpenAI has already confirmed a hefty purchase order! With low-cost inference solutions booming globally, the inference chip race is definitely getting intense! 🔥

8. Tencent internally tests QClaw integration with WeChat and QQ. Guess what? Tencent is currently internally testing its brand new QClaw tool! 🧪 This one-click installation and launch package (AI News) for the OpenClaw protocol means both WeChat and QQ can integrate and use it. Plus, it’s compatible with a bunch of top-tier large models! ✨ Seriously, the AI agent era is officially here!

AI News Daily: Multiple Channels

💬 WeChat Official Account	📹 Douyin
Official Account: Hexi2077	Self-media Account