11-21-Daily AI News Daily

AI News Daily 2025/11/21

AI News | Daily Brief | Web Data Aggregation | Cutting-Edge Science Exploration | Industry Voice | Open Source Innovation | AI & Human Future | Visit Web Version↗️ | Join Group Chat🤙

Today’s Summary

OpenAI releases its code model GPT-5.1-Codex-Max, breaking context window limits to compete with Google.
Quark AI Glasses partner deeply with Amap, enabling real-time navigation where arrows follow head movements.
Meta open-sources SAM 3D model, generating 3D objects and human bodies from single images, lowering creation barriers.
Embodied AI company Xingdong Jiyuan secures nearly 1 billion in funding; Musk and Nvidia plan a data center in Saudi Arabia.
ChatGPT launches a free teacher tool, while UK novelists worry about AI's impact on their copyrights and income.

Product & Feature Updates

1. Quark AI Glasses and Amap have leveled up their partnership, totally freeing up your hands and eyes! These smart glasses now support one-click navigation “screen mirroring” from your phone directly to the lenses. Even cooler, the navigation arrows follow your head movements in real-time, guiding you safely whether you’re out for a city walk or cycling – no more dangerous head-down phone glances. Plus, they’re set to integrate more ride-hailing and local lifestyle services, bringing a true “what you see is what you get” augmented reality experience. This is a total game-changer for anyone who gets lost easily! ✨

2. OpenAI has dropped a bombshell, launching GPT-5.1-Codex-Max, a new model specifically designed for code, making it clear they’re going head-to-head with Google – especially while everyone’s still buzzing about Gemini 3. The most insane feature of this new model is its breakthrough past context window limitations, allowing it to work continuously for over 24 hours like a tireless programmer. It also boasts higher inference efficiency, reducing thought token consumption by a whopping 30%. According to their Official Latest Blog (AI News) , the “endurance race” in AI programming is now in full swing, making marathon coding tasks totally possible! 🚀

3. Kling, the video generation tool, has also jumped into the “Black Friday” frenzy, offering creators an annual mega-deal! From November 17th to 28th, Kling is running its lowest price of the year with a 50% discount on first-time subscriptions and up to 40% cash back on recharges. Even more enticing, premium users can unlock unlimited generation mode during the event, letting their imaginations run wild without any limits. Click Here (AI News) to check out this unmissable creative spree! 🤯
   

4. OpenAI has officially rolled out ChatGPT for Teachers, crafting an exclusive “teaching superpower assistant” for K-12 educators in the US! This version not only offers practical features like lesson planning, question generation, and collaboration, but it also heavily emphasizes security controls compliant with the FERPA student privacy protection law, ensuring student data is never used for model training. Even better, this Teacher Edition Tool (AI News) will be freely available until June 2027, aiming to save teachers hours of work each week and truly make AI a powerful partner in the classroom. Talk about a bright idea! 🍎
   
   

Cutting-Edge Research

1. The DriveReg framework, introduced in a new paper, aims to make autonomous vehicles obey traffic rules just like seasoned drivers! It basically equips self-driving cars with a “legal advisor” fluent in local traffic regulations 🧠. This framework leverages RAG technology to retrieve relevant clauses from regulatory documents, then uses an LLM to reason and determine if an action is compliant, massively boosting decision transparency and trustworthiness. As This Cutting-Edge Research (AI News) shows, future smart cars won’t just need to drive well; they’ll need to be “licensed” and law-abiding too. 🚓

2. A new paper’s FinCriticalED visual benchmark is shaking up how we evaluate the accuracy of financial document OCR – turns out, just checking if the text is literally correct isn’t enough! This is the first time evaluation standards have jumped from “literal similarity” to “factual correctness.” 🎉 Through expert annotations, FinCriticalED specifically examines models’ understanding of critical financial facts like positive/negative signs for amounts and dates, acting like a “financial literacy exam” for visual language models. View This Research (AI News) to see that even the strongest models still stumble when faced with complex financial tables. Ouch! 💸

Industry Outlook & Social Impact

1. Xiaomi Auto has hit a major milestone, with its 500,000th vehicle rolling off the production line! Founder Lei Jun excitedly declared this just a new beginning. He solemnly pledged to focus on three key areas going forward: vigorously prioritizing safety and delivery, boosting technological innovation, and actively pushing for the deep integration of AI and smart manufacturing. As This Report (AI News) emphasizes, Xiaomi isn’t just building cars; it’s blazing a new trail for AI-powered high-end manufacturing. The future looks super promising! ✨

2. UK novelists are getting seriously worried as generative AI rises, with over half fearing their livelihoods are at risk. A Cambridge University report revealed that nearly two-thirds of writers claim their work has been used without permission to train large models, and a staggering 39% have already seen their income impacted. While the outlook seems grim, as This Survey Report (AI News) shows, this high-stakes game about copyright, originality, and the future has only just begun. Grab your popcorn! 🍿

3. Embodied AI company Xingdong Jiyuan is shaking up the scene once again, announcing it has secured nearly 1 billion RMB in A+ round financing, led by Geely Capital, with total orders already exceeding 500 million RMB! 💸 This company is committed to full-stack R&D, integrating its “embodied brain” ERA-42 with its “humanoid body” hardware and software. They’ve already achieved commercial implementation globally, boasting 9 of the world’s TOP 10 tech giants as clients. As This In-Depth Report (AI News) explains, this funding will supercharge their “model-body-data” evolutionary flywheel, bringing us closer to an era of general-purpose robot companions. How cool is that?! 🤖

4. Elon Musk and Nvidia CEO Jensen Huang have jointly announced plans to team up with Saudi Arabia’s Humain AI company to build a colossal 500-megawatt data center in Saudi Arabia, accelerating the marriage of tech giants and Middle Eastern capital! 🤯 This “behemoth” facility will be entirely powered by Nvidia computing chips, providing serious horsepower for Musk’s xAI. As This Major News (AI News) hints, behind the global AI computing race lies a high-stakes game of capital, technology, and geopolitics. Get ready for some fireworks! 💥

5. As emotional bonds with AI companions deepen, industry giants are finally getting serious about building healthy human-AI relationships. 🤔 A post-Stanford workshop insight points out that safety guidelines just aren’t enough; AI companions need a true technical architecture, including contextual memory, emotional trajectory modeling, and dependency detection. As This Developer’s Share (AI News) explains, we can’t just patch up “architectural flaws” with “policy fixes.” Building AI with a “soul” needs to start from the foundational logic. A real brain-buster! 🧠
   

Open Source TOP Projects

1. Meta AI has once again rocked the visual field, open-sourcing the groundbreaking SAM 3D model, making 3D model generation from a single image a reality! 🔥 This project includes two core components: SAM 3D Objects (for reconstructing objects) and SAM 3D Body (for reconstructing human bodies). It can accurately restore the 3D shape, texture, and even the full pose of human bodies from any everyday photo. As demonstrated in this Detailed Technical Interpretation (AI News) and the Official Project Page (AI News) , this isn’t just a technological leap; it’s lowering the barrier to 3D creation to an unprecedented degree. Get ready for an era where “everything can be 3D”! 🚀
   

Social Media Shares

1. The Google developer community is enthusiastically showcasing the amazing potential of Gemini 3 in game development – turns out, anyone can now create 3D interactive games! 🎉 With Gemini 3, you can build fully interactive, well-designed 3D games, completely breaking down the barriers between creativity and technology. Go Check Out These Community Works (AI News) for inspiration – who knows, the next big hit game might just come from you! 🎮

2. Want to jump right into Gemini 3’s powerful capabilities? Google has already prepped a quick-start guide just for you! 🤩 Through a short video tutorial, developers can learn how to create API keys, send their first request, and leverage multi-tool capabilities (like search and code execution). What are you waiting for? Come Watch This Introductory Video (AI News) and kick off your Gemini 3 development journey! 🚀
   

3. Good news for anime and manga fans! A tool called Nano Banana Pro is blowing up on social media because it can complete the entire process of manga translation, super-resolution upscaling, and coloring with just one click! 🤯 This means manga lovers no longer have to painstakingly wait for fan translation groups; they can enjoy high-definition, full-color reading instantly. As This Netizen’s Excited Share (AI News) perfectly put it, this is an absolute “divine tool” custom-made for manga enthusiasts. Seriously cool! ✨
   

4. How can multi-agent systems escape the “talking past each other” dilemma? A new study suggests the key is for AI to learn to “speak,” rather than just “guessing” through actions! 🗣️ This framework introduces language-enhanced multi-agent reinforcement learning, enabling agents to clearly articulate strategies and negotiate using natural language during decision-making. As This Tweet’s Interpretation (AI News) explains, this makes AI behavior explainable and auditable, a crucial step towards truly intelligent collaboration. Pretty insightful, huh?💡
   

5. The “battle of the AI gods” has spilled over onto Wall Street, with a unique US stock trading competition now underway! 🤑 The newly released Gemini 3 has parachuted into the arena, competing against top AI models like GPT-5.1 and DeepSeek to see who can rake in the highest investment returns. As The Promotion of This Competition (AI News) points out, with AI getting into stock trading, the human world is getting wilder by the minute! Come vote for your favorite AI champion. Who’s gonna win?! 🤔
   

6. How powerful is Nano Banana Pro, really? A simple, direct comparison image says it all! ✨ Compared to the standard version, the Pro version’s generated images show a visibly huge leap in detail, texture, and lighting – it’s like the characters have been given souls. As This User’s Intuitive Demonstration (AI News) clearly shows, this isn’t just a simple upgrade; it’s a complete evolution! 🔥
   
   

7. Talk about creativity! A user has already cooked up a brilliant combo for SAM 3D and Nano Banana, allowing for custom character generation with total freedom! 💡 The workflow involves using SAM 3D to batch convert real-person action videos into 3D puppet poses, then taking screenshots to feed into Nano Banana as pose reference images. This lets you generate specific characters from any angle and with any action. As This Creator’s Vision (AI News) shows, combining AI tools is unlocking unprecedented creative freedom. Mind blown! 🤯
   

8. Sometimes, what stops companies from fully embracing AI isn’t the tech or the budget, but a tiny little invoice! 🧾 A netizen posed a soul-searching question: “How does your company get invoices for Figma or ChatGPT purchases?” This sparked widespread agreement. As This Complaint (AI News) reveals, these seemingly trivial procedural details are often the most frustrating “last mile” problems when advanced foreign tools try to land in domestic companies. Talk about a headache! 😵‍💫

AI News Daily Audio Version

🎙️ Xiaoyuzhou	📹 Douyin
Laisheng Xiaojiuguan	Self-Media Account