Samsung is launching a company-wide AI initiative aimed at integrating artificial intelligence into daily operations across its businesses.

Announced on June 9, 2026, the program, called “AI Transformation” (AX), will introduce new AI tools, employee training programs, and dedicated AI teams throughout the Samsung Group. The effort is designed to increase productivity and encourage wider use of AI across the company’s operations.

As part of the rollout, Samsung plans to give employees access to external generative AI services, including Google Gemini, ChatGPT, and Claude. The tools will be available across subsidiaries and are expected to support tasks ranging from software development and marketing to manufacturing and business operations.

Training is another major part of the initiative. Samsung aims to provide AI education to its entire workforce of roughly 280,000 employees by the end of 2026.

The company is also starting from the top. Around 50 group presidents will attend a two-day AX Bootcamp at Samsung’s Human Resources Development Institute this month. Senior executives, numbering about 2,300, will take part in separate three-day training sessions, with the first round scheduled to finish by mid-August.

Samsung is expected to unveil a group-wide AX Vision during the leadership training program, outlining how AI will be applied across its businesses. The company also plans to establish dedicated AI departments within each subsidiary to oversee adoption and develop internal expertise.

The move reflects the growing importance of AI across the technology industry, where companies are increasingly looking to automate workflows and improve decision-making. This is one of Samsung’s largest efforts yet to expand AI adoption beyond products and into the way the company itself operates.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source: Chosunbiz)

The post Samsung launches company-wide “AI Transformation” initiative to increase productivity appeared first on Gizmochina.

NVIDIA’s new RTX Spark platform was one of the biggest highlights at Computex Taipei 2026, with major PC brands including ASUS, Dell, Lenovo, and MSI unveiling compact desktop systems powered by the new AI-focused architecture. The launch marks NVIDIA’s first major step into the PC processor market and could reshape the future of AI PCs, creator workstations, and compact desktops.

NVIDIA RTX Spark Platform Takes Center Stage

The RTX Spark platform is an Arm-based system-on-chip (SoC) built around NVIDIA’s Grace and Blackwell technologies. It combines a 20-core CPU with a Blackwell GPU featuring 48 SM units and 6,144 CUDA cores. The platform also supports up to 128GB of LPDDR5X unified memory.

NVIDIA says the chip is designed for local AI development, content creation, gaming, and AI inference workloads. The company also claims graphics performance comparable to an RTX 5070-class GPU in certain scenarios.

ASUS, Dell, Lenovo, and MSI Showcase New Mini PCs

Several manufacturers introduced RTX Spark-powered mini PCs during the event. While each system features a unique chassis design, they share the same core platform and connectivity options.

ASUS ProArt GA10

The ASUS ProArt GA10 features a compact 150 × 150 × 51 mm design and supports up to 140W of cooling capacity. It includes four USB-C ports, HDMI output, and 10 Gigabit Ethernet connectivity.

Dell XPS RTX Spark Desktop

Dell’s new compact desktop also adopts the RTX Spark platform. The system includes four USB-C ports, HDMI output, and 10GbE networking while maintaining a small footprint.

Lenovo RTX Spark Mini PC

Lenovo introduced a newly designed mini PC chassis built around RTX Spark. It offers the same set of connectivity options, including four USB-C ports, HDMI, and 10GbE Ethernet.

MSI EdgeMesa N AI+

MSI’s EdgeMesa N AI+ stands out with its white chassis design. The system targets content creators and AI developers while offering the same RTX Spark hardware foundation.

Shared Features Across All Models

Every RTX Spark mini PC showcased at Computex includes 10GbE networking, 20Gbps USB-C connectivity, PCIe Gen 5 M.2 SSD support, and up to 128GB unified memory.

Why RTX Spark Could Be a Game-Changer for PC Buyers

For consumers, RTX Spark could bring desktop-class AI performance, powerful graphics, and massive memory capacities into much smaller and more efficient PCs. Because the CPU, GPU, and up to 128GB of unified memory work together on a single platform, users may be able to run advanced AI models, edit high-resolution videos, and handle demanding creative workloads locally without relying heavily on cloud services.

If NVIDIA’s performance claims hold up, RTX Spark systems could also offer a compelling alternative to traditional Intel and AMD-powered PCs, increasing competition and giving buyers more choice. 

The trade-off is that some older Windows applications and specialized software may face compatibility or optimization challenges, although Windows on Arm support has improved significantly in recent years.

Read More:

• US- China Tech War: China Just Beat NVIDIA in the Next Big AI Race
• Why Nvidia’s H2+ Humanoid Robot Has the Tech World Talking
• US Expands AI Chip Restrictions Worldwide: What It Means for China

(via)

The post NVIDIA RTX Spark Mini PCs Arrive at Computex 2026 — Could They Replace Traditional Desktops? appeared first on Gizmochina.

The battle for AI leadership is rapidly moving from chatbots to robots, and China has just scored a major victory. Chinese startup Spirit AI announced that its Spirit v1.6 model ranked No.1 on the RoboArena benchmark, outperforming NVIDIA and other leading competitors. 

Spirit v1.6 achieved a score of 1,924, outperforming NVIDIA’s Cosmos3-Nano-Policy, which scored 1,881, and DreamZero, another NVIDIA-backed model, which scored 1,763. The achievement marks the first time a Chinese embodied AI model has reached the No.1 spot on the benchmark.

Why RoboArena Matters

RoboArena is designed to measure how effectively AI systems can translate decisions into real-world robotic actions. Unlike traditional AI benchmarks that focus on language or reasoning, RoboArena evaluates physical task execution, including object manipulation, navigation, tool usage, perception, planning, and adaptability in unfamiliar environments.

The benchmark is widely respected because it uses randomized environments, adversarial testing scenarios, and strict anti-overfitting measures, making it difficult for companies to optimize specifically for the test.

The Growing Physical AI Race

The result highlights how competition in artificial intelligence is increasingly shifting beyond chatbots and large language models toward physical AI. These systems serve as the “brains” behind humanoid robots, robotic arms, autonomous vehicles, and industrial automation platforms.

Spirit AI’s success is being viewed as evidence that efficient model design and high-quality training data can be just as important as massive computing power. The company’s model reportedly demonstrated strong performance in object handling, autonomous navigation, complex tool use, and adapting to new situations with minimal fine-tuning.

China’s Expanding Advantage

China has also been gaining ground across several other embodied AI benchmarks. Companies including Manifold AI, AgiBot, and DexForce currently lead categories related to world modeling, robotic perception, and robotics data generation.

At the same time, investment in the sector continues to surge. Spirit AI recently raised 1.5 billion yuan (about $222 million), while other Chinese robotics startups have secured large funding rounds over the past year.  China’s growing strength in robotics data, funding, and benchmark performance suggests the global race is becoming more competitive than ever.

Read More:

• ASUS Launches ROG Harpe II Extreme Edition 20 Gaming Mouse With 24K Gold-Plated Design
• Why Nvidia’s H2+ Humanoid Robot Has the Tech World Talking
• US Expands AI Chip Restrictions Worldwide: What It Means for China

(via)

The post US- China Tech War: China Just Beat NVIDIA in the Next Big AI Race appeared first on Gizmochina.

Nvidia has announced a major step forward in humanoid robotics with the launch of the H2+ reference robot platform, developed in partnership with China’s Unitree Robotics and Singapore-based Sharpa. The announcement was made by Nvidia CEO Jensen Huang during his Computex 2026 keynote in Taipei.

The project is attracting industry attention because it combines advanced AI, robotics hardware, and software into a unified platform designed to speed up the development of real-world humanoid robots. 

Rather than being a finished commercial robot, H2+ is a reference design that researchers, universities, and robotics companies can use as a blueprint to build and customize their own humanoid robots. Nvidia says the platform is designed to simplify every stage of robot development, from data collection and AI training to real-world deployment.

Combining Hardware and AI

The H2+ platform brings together several technologies into a single package. It uses Unitree’s H2 Plus humanoid robot body, Sharpa’s Wave robotic hands, and Nvidia’s Isaac GR00T AI models. Powering the system is Nvidia’s new Jetson AGX Thor T5000 computer, which is based on the Blackwell GPU architecture.

The robot stands nearly six feet tall and offers a total of 75 degrees of freedom, including highly dexterous robotic hands with 22 active joints. Nvidia says the Jetson AGX Thor T5000 delivers up to 2,070 FP4 TOPS of AI performance and includes 128GB of memory, allowing the robot to process complex tasks in real time.

Nvidia Introduces Cosmos 3

Alongside H2+, Nvidia also introduced Cosmos 3, a new physical AI world model designed to help robots better understand and learn from their environments.

According to Huang, one of the biggest challenges in robotics is collecting enough useful training data. Cosmos 3 can understand both third-person and first-person perspectives, helping generate robot-centric training data that can improve learning and decision-making.

Expansion Beyond China

Nvidia also revealed plans to expand its humanoid robot partnerships beyond China. The company is looking to work with robot makers in the United States, Europe, and South Korea, although partner names have not yet been announced.

The goal is to create standardized robot platforms with built-in security features such as Secure Boot and Confidential Computing, helping ensure that only verified software can run on the machines.

Growing Humanoid Robot Market

The announcement comes as interest in humanoid robots continues to grow. Analysts at Morgan Stanley expect China’s humanoid robot sales to exceed 28,000 units in 2026, potentially making it the world’s largest market for the technology.

With H2+, Cosmos 3, and new international partnerships, Nvidia is positioning itself at the center of the rapidly expanding humanoid robotics industry, supplying the AI hardware, software, and development tools that could power the next generation of intelligent machines.

Read More:

• US Expands AI Chip Restrictions Worldwide: What It Means for China
• Xiaomi is repairing its rainstorm-affected phones, laptops, and accessories for free in China
• China vs US in Robotics: Nvidia CEO Reveals the Real Power Shift

(via)

The post Why Nvidia’s H2+ Humanoid Robot Has the Tech World Talking appeared first on Gizmochina.

The US government has given a new clarification that export restrictions on advanced AI chips apply not only to companies operating in China but also to Chinese-owned firms and subsidiaries located overseas. The move closes a potential loophole that may have allowed Chinese companies to access restricted AI hardware through operations in countries outside mainland China.

Why the US Made This Move

The clarification follows concerns in Washington that Chinese firms could obtain advanced AI processors through overseas subsidiaries despite existing export controls. Some policymakers argued that uncertainty created after the cancellation of the Biden-era AI Diffusion Framework may have opened a pathway for such purchases. The new guidance makes it clear that Chinese-headquartered companies remain subject to licensing requirements regardless of where they operate.

Pressure Grows on China’s AI Industry

The tighter enforcement is expected to make access to advanced US AI hardware more difficult for Chinese companies. NVIDIA’s most powerful Blackwell AI chips remain restricted, while purchases of other advanced processors continue to face regulatory scrutiny. As a result, Chinese technology firms are being pushed to rely more heavily on domestic alternatives.

China Accelerates Its Chip Strategy

In response to years of US restrictions, China has been rapidly expanding its domestic AI semiconductor industry. Companies including Huawei, Moore Threads, Biren Technology, Cambricon, and Alibaba have increased investments in AI processors designed for both training and deploying large AI models. Rather than focusing solely on replacing Nvidia products, many firms are now building complete AI hardware and software ecosystems.

Domestic Alternatives Gain Importance

Industry analysts expect Chinese AI companies to increasingly adopt locally developed accelerators as access to foreign chips becomes more limited. Huawei’s Ascend series and Alibaba’s custom AI processors are among the products attracting attention as China works to reduce dependence on US technology.

What to Expect Next

The latest US action is unlikely to slow China’s AI ambitions. Instead, it may further accelerate investment in domestic chips, software platforms, and AI infrastructure. The competition between the US and China is increasingly shifting from individual products to entire technology ecosystems. In the coming years, the key question may no longer be whether China can replace Nvidia, but how quickly it can build a self-sufficient AI industry capable of competing on a global scale.

Read More:

• China vs US in Robotics: Nvidia CEO Reveals the Real Power Shift
• Huawei aims to launch lithography equivalent to TSMC’s 1.4nm by 2031 in a major defiance of U.S. sanctions
• OnePlus Ace 6 Ultra vs OnePlus Ace 6: Are the Upgrades Really Worth the Extra Money?

The post US Expands AI Chip Restrictions Worldwide: What It Means for China appeared first on Gizmochina.

CHUWI has officially introduced the new CoreBook Air, a next-generation Copilot+ PC designed to deliver powerful AI capabilities, long battery life, and premium portability. According to CHUWI CEO Jesea Yang, the CoreBook Air combines Lunar Lake performance and Copilot+ AI capabilities in an ultra-light 1.2kg design, helping bring advanced AI PCs to a wider audience at a more accessible price point. The company aims to make advanced AI-powered computing more accessible to students, professionals, creators, and remote workers.

Premium Design Built for Portability

The CHUWI CoreBook Air features a slim and lightweight design that weighs just 1.2kg, making it easy to carry throughout the day. The laptop uses a premium CNC-crafted aluminum chassis that provides a durable and professional appearance while maintaining portability. Its compact form factor makes it suitable for students attending classes, content creators working on the move, business professionals, and remote workers who need a dependable travel companion.

Sharp 2.8K Display with Smooth Visuals

The laptop comes equipped with a 14-inch IPS display featuring a 2.8K resolution of 2880 × 1800 pixels. It also supports a 120Hz refresh rate and covers 100% of the sRGB color gamut. This combination delivers sharper text, detailed visuals, smoother scrolling, and an improved video viewing experience. The high-resolution panel is also designed to provide comfortable viewing during long work sessions.

Intel Lunar Lake Performance with Advanced AI

At the heart of the CoreBook Air is the Intel Core Ultra 5 226V processor based on Intel’s Lunar Lake (Series 2) architecture. The chip features 8 CPU cores and 8 threads, with boost frequencies reaching up to 4.5GHz. Users can expect strong computing performance, improved power efficiency, broad software compatibility, enhanced security, and advanced AI processing capabilities.

Graphics duties are handled by the integrated Intel Arc 130V GPU, which CHUWI claims can deliver up to twice the graphics performance of older Intel Iris Xe solutions. The GPU supports hardware-accelerated ray tracing, casual gaming, content creation tasks, multimedia streaming, and GPU-accelerated applications.

Copilot+ Features and AI Acceleration

One of the biggest highlights of the CoreBook Air is its AI performance. The system delivers up to 97 TOPS of total AI computing power and includes an Intel AI Boost NPU capable of up to 40 NPU TOPS. This dedicated AI hardware enables faster on-device AI processing, reduced dependence on cloud services, improved privacy, and quicker AI responses.

Running Windows 11 Pro out of the box, the laptop supports Microsoft Copilot+ PC experiences such as Click to Do, Cocreator, and Live Captions. A dedicated AI key also provides instant access to Microsoft Copilot.

Memory, Storage, Battery, and Connectivity

The CoreBook Air includes 16GB LPDDR5X memory running at 8533 MT/s using a memory-on-package design for faster multitasking, lower latency, and better efficiency. Storage is handled by a 512GB PCIe SSD, while users can upgrade storage later through PCIe 4.0 x4 SSD support.

A 55Wh battery powers the laptop and is rated for up to 15 hours of local video playback. Charging is handled by a 65W USB-C Power Delivery adapter, capable of charging the battery to around 63% in approximately one hour.

For connectivity, the laptop supports Wi-Fi 6E for faster wireless speeds, lower latency, and improved stability. Port selection includes two Thunderbolt 4 ports with up to 40Gbps bandwidth, Power Delivery support, external display connectivity, and eGPU compatibility. Users also get three USB-A ports, an HDMI 2.0 output, and a 3.5mm audio jack.

Pricing and Availability

The CHUWI CoreBook Air is available for pre-order now. The laptop carries an official retail price of $799, but early buyers can take advantage of a limited-time Super Early-Bird discount of $50, reducing the launch price to $749. The promotional offer is available on a first-come, first-served basis while launch stock lasts.

CHUWI CoreBook Air – Specs Sheet

Display

Performance

Memory & Storage

Battery & Charging

AI Features

Ports & Connectivity

Wireless Connectivity

Camera & Multimedia

The post CHUWI CoreBook Air Launches as a Copilot+ PC with Intel Lunar Lake, 2.8K Display and 15-Hour Battery appeared first on Gizmochina.

Xiaomi has introduced its latest open-ear audio wearable, the Xiaomi Clip Earbuds, combining lightweight comfort, powerful sound, AI real-time translation, and long battery life in a stylish clip-on design. The new earbuds are now available on Giztop for a starting price of $159.99 with global shipping support. 

Lightweight and Comfortable Design

The Xiaomi Clip Earbuds are designed for users who want comfort during long listening sessions. The earbuds feature a lightweight clip-on design, with each earbud weighing around 5.5 grams. The open-ear style allows users to stay aware of their surroundings while listening to music or taking calls. This design makes them suitable for workouts, office use, travel, and daily outdoor activities.

The earbuds also come with an IP57 dust and water resistance rating, offering protection against sweat and light rain. Xiaomi has included touch controls on the earbuds for easy music playback, call handling, and other functions. MIUI users also get a pop-up window for quick connection and battery status viewing.

Powerful Audio and Clear Calling Experience

Xiaomi Clip Earbuds are equipped with 11mm dynamic drivers paired with microcrystalline metal-coated diaphragms. The earbuds deliver balanced sound with clear vocals and deep bass. They support a wide 20Hz to 48kHz frequency response for detailed audio performance.

For wireless audio, the earbuds support AAC, SBC, and LHDC 5.0 codecs. They also carry Hi-Res Audio Wireless certification, helping deliver high-quality sound during music playback.

Call quality is another major highlight. Xiaomi uses a three-microphone setup along with a VPU bone-conduction microphone to reduce surrounding noise and isolate the user’s voice. This helps provide clearer conversations even in crowded or noisy environments.

Long Battery Life and Smart Features

The Xiaomi Clip Earbuds offer up to 9 hours of music playback on a single charge. With the charging case, total battery life can reach up to 38 hours. The USB-C charging case also supports fast charging, and a quick 5-minute charge can provide around 2 hours of playback.

Another interesting feature is AI real-time translation support, making the earbuds more useful for travel and communication.

With premium audio features, clear call quality, smart AI tools, and a comfortable open-ear design, the Xiaomi Clip Earbuds offer a modern listening experience for everyday users. 

Read More:

• Xiaomi 17 Max Now Available on Giztop With 8000mAh Battery and 200MP Leica Camera
• Lenovo Legion Y900 13-Inch 2026 Now Available at Giztop Starting at $499
• Lenovo’s New Legion Y700 2026 Gaming Tablet Gets Massive $200 Discount at Giztop

The post Xiaomi Clip Earbuds Now Available on Giztop for $159.99 – Open-Ear Design Meets AI Translation appeared first on Gizmochina.

China has officially launched its first general-purpose household robot, called “Shiguang S1,” in Wuhan’s Optics Valley. The robot was introduced on May 20 by Chinese robotics company GigaBrain and is designed specifically for daily home use. Unlike traditional robots built for factories, the Shiguang S1 is made to assist families with regular household tasks and interact naturally with people inside the home.

Designed to Handle Daily Household Work

The Shiguang S1 can perform several common household activities. According to the company, the robot is capable of folding clothes, cooking meals, and clearing the dining table. It can also chat with elderly people and provide assistance at home. One of its biggest features is its ability to continuously learn and adapt to new tasks, making it more flexible than older task-specific robots.

Powered by Advanced AI Technology

GigaBrain says the robot uses its self-developed embodied intelligent model, which allows it to understand tasks and react to different home environments. Ye Yun, Vice President of R&D at GigaBrain, explained that traditional factory robots usually follow fixed instructions, while the Shiguang S1 can plan actions on its own and respond intelligently to changing situations.

The company also revealed plans to officially launch its next-generation “GigaBrain 1” model in the third quarter of 2026.

Safety and Early User Testing

To improve safety, the robot includes a compliant control mechanism. If it accidentally touches a person or pet while moving, it immediately stops to reduce the risk of injury.

Families in Wuhan can now apply to become early “seed users” through the “Shiguang Vision” WeChat account and the company’s official application page. The launch highlights China’s growing progress in AI-powered robotics, elderly care technology, and smart home automation.

Read More:

• Motorola Edge 70 Pro vs Redmi Note 15 Pro+: Which Mid-Range Phone Is Better
• OPPO Find X9 Ultra vs Honor Magic8 Pro Comparison: Camera, Battery & Performance
• Motorola Razr Ultra 2026 vs Honor Magic V5: One Clearly Offers More Value

(via)

The post China Launches First Universal Household Robot “Shiguang S1” in Wuhan appeared first on Gizmochina.

Kynooe has officially launched its new AI-powered robotic arm on Kickstarter, introducing a fully modular robotic arm designed for everyday users rather than just engineers or robotics experts.

Built for home assistance, smart automation, content creation, and creative workflows, Kynooe combines modular hardware, AI-powered interaction, and app-based controls into a more consumer-friendly robotics platform. The project focuses on making robotics more accessible through a no-code experience and a flexible modular ecosystem.

Beginner-Friendly Setup Without Coding

One of the key highlights of Kynooe is its beginner-focused setup process. The system uses a plug-and-play modular design and can be controlled through a companion app with an interface designed to feel as intuitive as using a smartphone.

Instead of relying on traditional robotics programming or complex engineering setups, users can control and customize the robot through visual interface tools, AI-assisted interactions, and simplified workflows. The no-code approach is designed to lower the barrier to entry for users who are new to robotics.

Fully Modular Design With Expandable Accessories

Unlike traditional robotic arms built for fixed functions, Kynooe features a fully modular architecture that supports interchangeable tools, accessories, and attachments. The platform is compatible with precision instruments, custom modules, and even 3D-printed extensions for specialized use cases.

The company sums up the concept with the tagline “One arm. Infinite applications,” emphasizing the system’s adaptability across diverse workflows and environments. Depending on the configuration, the robot can shift seamlessly between creative assistance, intelligent filming, automation tasks, and productivity-focused applications.

AI Features and Natural Interaction

Kynooe also integrates AI-powered interaction features aimed at making the experience feel more conversational and intuitive. Users can interact with the robot through natural language commands instead of relying entirely on manual controls or traditional programming methods.

One of the major built-in features is on-device AI face tracking. The feature is aimed at content creators and smart filming setups. The robotic arm can automatically track faces and adjust camera movement for smoother recording and hands-free filming support. The company also plans to introduce additional on-device AI features in the future, aiming to improve user experience while maintaining user privacy.

Built for Everyday Scenarios

Rather than focusing solely on industrial or engineering applications, Kynooe is positioned as a robotics platform for real-world, everyday use cases. The system is designed for household assistance, entertainment, smart filming, creative projects, and interactive experiences with family and friends.

The robotic arm adapts across different scenarios depending on the selected modules and accessories, allowing users to customize the system around their own workflows and creative needs.

Open Ecosystem and Kynooe Robot Hub

Beyond hardware, Kynooe is also building an ecosystem focused on embodied AI and community-driven robotics development. The company plans to release an open-source SDK on GitHub that will provide developers and advanced users with access to everything from AI integrations to low-level robotic joint controls.

The ecosystem also includes the Kynooe Robot Hub, a platform where users can train, customize, publish, and share robotic behaviors and workflows with the broader community. The company envisions a future where robotic skills can be shared, downloaded, and deployed as easily as mobile apps.

Development Progress and Manufacturing Timeline

According to the company, Kynooe has already completed structural design work, hardware validation, and small-batch trial production. The project is currently moving toward certification, production validation testing, and mass production preparation ahead of planned shipments later in 2026.

The current roadmap includes ongoing tooling optimization, manufacturing preparation, and compliance testing as the company scales toward full production.

Shipping Plans and Kickstarter Offers

Kynooe is scheduled to officially launch on Kickstarter on May 20. The campaign will feature a range of launch-day discounts, limited-edition rewards, and multiple pledge options designed for different types of users, including creator-focused bundles and modular accessory packages.

According to the company, the first batch of shipments is currently expected to begin around September 2026, followed by ongoing fulfillment and second-batch deliveries planned for October 2026.

More information about the project is available on the official Kickstarter campaign page.

Verdict

Rather than building yet another industrial robotic arm, Kynooe aims to reimagine robotics as a flexible everyday companion for creators, smart homes, and personal automation. Its combination of modular hardware, AI-powered interaction, and beginner-friendly controls makes it one of the most ambitious consumer robotics projects currently entering the crowdfunding space.

The post Kynooe Launches World’s First Fully Modular AI-Powered Robotic Arm on Kickstarter appeared first on Gizmochina.

Google says it has identified what may be the first real-world zero-day exploit developed with the help of artificial intelligence, marking a potentially important shift in how cyberattacks are being created.

In a report published by Google’s Threat Intelligence Group (GTIG) on May 11, the company said attackers used AI assistance to discover and exploit a previously unknown vulnerability in a widely used open-source web administration tool. The flaw reportedly allowed attackers to bypass two-factor authentication after obtaining login credentials.

According to Google, the vulnerability was disclosed to the affected vendor and patched before it could be used in larger attacks.

What caught researchers’ attention was the exploit code itself. Google says the Python script showed several signs typically associated with AI-generated code, including overly instructional comments, structured textbook-style formatting, and even a hallucinated CVSS severity score that did not actually exist.

Interestingly, the vulnerability was not a traditional memory corruption bug or a simple coding mistake. Instead, it involved higher-level business logic and semantic analysis, an area where modern AI models have started showing stronger capabilities. In simple terms, the AI appears to have helped connect pieces of application behavior that could eventually be abused together.

Security researchers have been warning for a while that AI could lower the barrier for more advanced cyberattacks, especially by helping less experienced attackers automate parts of the process that previously required deeper technical expertise.

At the same time, companies like Google are also using AI defensively for things like automated bug discovery, threat analysis, and patch development. Still, this latest case is likely to add more urgency to conversations around AI-powered cybersecurity threats.

While the incident itself appears to have been contained before wider abuse, researchers see it as an early sign of how AI may increasingly become part of both sides of the cybersecurity landscape.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source: Google)

The post Google detects first known AI-assisted zero-day exploit in the wild appeared first on Gizmochina.

AI video creation is becoming faster and smarter, but most workflows still require multiple tools and unpredictable costs. ShengShu Technology now aims to change that with Vidu Claw, a new AI-powered platform that can turn a simple text prompt into a fully finished advertisement in minutes. 

What Is Vidu Claw?

ShengShu Technology officially introduced Vidu Claw in Singapore on May 8, 2026. The platform is being described as the world’s first “AI CMO,” short for AI Chief Marketing Officer. Its main goal is simple: turn a short text prompt into a fully finished advertisement without requiring multiple tools or production teams.

Vidu Claw is designed as a complete AI-powered marketing and video production platform. Instead of using separate apps for scriptwriting, editing, voiceovers, and animation, the system handles the entire process automatically.

Solving a Major AI Video Industry Problem

According to ShengShu, today’s AI video workflows are still too fragmented and expensive. Most creators and brands need different platforms for writing scripts, creating storyboards, generating visuals, editing videos, and adding music or voiceovers. On top of that, many AI video platforms depend heavily on third-party AI models and credit-based pricing systems.

ShengShu argues that this creates what it calls an “aggregator tax,” where users pay unpredictable costs for generations, revisions, and failed outputs. The company says Vidu Claw was built to simplify the workflow while reducing both production time and costs.

How Vidu Claw Works

The biggest highlight of Vidu Claw is its fully automated workflow. A user can enter a simple prompt such as:

“Create a short-form ad for a running shoe targeting young women on Instagram.”

The system then automatically handles creative concept generation, scriptwriting, storyboarding, visual creation, voiceovers, background music, and final video assembly. The result is a ready-to-publish advertisement optimized for social media platforms.

ShengShu says the platform acts like an autonomous marketing strategist rather than just a video generator.

Technology Behind the Platform

Vidu Claw runs on ShengShu’s proprietary Vidu Q3 video model and is built on the open-source OpenClaw framework. The platform also includes an AI orchestration layer that coordinates every stage of production automatically.

Unlike many competitors, ShengShu claims Vidu Claw does not rely on external AI video engines for generation. Every frame is produced directly through Vidu Q3. The company says this improves visual consistency, speeds up rendering, and lowers operational costs.

Pricing and Enterprise Features

Instead of charging users per AI generation or revision, ShengShu is introducing a “Video Plan” subscription system. Customers pay for completed advertisements rather than individual credits. The company believes this will make budgeting easier for businesses and reduce financial risk during large campaigns.

Vidu Claw also includes a large “Vidu Skills” ecosystem with tools for AI avatars, faceless videos, social media automation, and marketing content creation. Because the platform is based on OpenClaw architecture, enterprises and developers can integrate it into existing marketing systems and AI agent frameworks.

ShengShu’s Bigger Vision

ShengShu believes the future of advertising is not just about faster editing tools but fully autonomous marketing infrastructure. The company claims Vidu Claw can reduce production costs to around 1% of traditional advertising budgets while compressing multi-day workflows into same-day delivery.

Beyond video creation, the AI CMO system can also assist with trend analysis, copywriting, reporting, analytics, presentation creation, and performance tracking. With Vidu Claw, ShengShu is positioning itself at the center of the growing competition in autonomous AI marketing technology.

The post Vidu Claw AI Can Turn a Simple Text Prompt Into a Complete Advertisement appeared first on Gizmochina.

Google just announced a new set of AI tools for Android called Gemini Intelligence. The basic idea is to have your phone handle more of your daily busywork. These features are rolling out this summer, starting with the new Samsung Galaxy S26 and Google Pixel 10, and will eventually come to other Android phones, smartwatches, and laptops later in the year.

The main change is app automation. Instead of just answering questions, Gemini can now navigate your phone to do things like book a workout class or order groceries. You can even long-press the power button over a screenshot of a grocery list, and Gemini will build the cart for you. You can watch its progress through a notification and just confirm the final purchase.

Browsing is getting a similar update. In late June, Chrome for Android will add an auto browse tool to help with routine tasks like reserving parking. Android’s Autofill is also getting a boost, pulling context from your apps to fill out complex forms so you don’t have to type as much.

For texting, Gboard is adding a voice feature called Rambler. If you tend to use filler words, pause, or switch languages while using voice dictation, Rambler will clean up the audio and turn it into a clear text message before you send it.

Google is also adding a text-to-widget tool. You can type a prompt, like asking for a weekly recipe list or a widget that only shows wind speed, and Android will build a custom widget for your home screen.

Because giving AI access to your screen requires trust, Google is keeping these features opt-in. The company says it uses secure environments to process the data, and an upcoming update to the Android Privacy Dashboard will let you see exactly which apps the AI interacted with over the past 24 hours.

And whenever the assistant is doing something in the background, a notification will stay at the top of your screen that you can’t swipe away, so you’ll always know when it’s running.

For more daily updates, please visit our News Section.

Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories!

(Google)

The post Google is turning Android into an AI agent with Gemini Intelligence appeared first on Gizmochina.

Alibaba has announced a major software upgrade for the Qwen AI Glasses S1, bringing a more proactive approach to AI interactions. Instead of waiting for voice commands, the smart glasses can now offer contextual suggestions and reminders based on a user’s schedule, surroundings, and daily habits. The update also introduces spatial 3D display technology and several lifestyle-focused AI services.

Qwen AI Glasses S1 now acts before you ask

The latest update allows the Qwen AI Glasses S1 to provide real-time prompts without requiring users to manually trigger commands. For example, the glasses can remind users to carry an umbrella before leaving home if rain is expected. During work hours, the device can also suggest taking a short break or stretching based on posture, wear duration, and activity patterns.

Alibaba says the AI system combines factors such as location, time, weather conditions, and usage behaviour to generate contextual recommendations. The company plans to further expand these abilities through stronger multimodal understanding and memory features. Future updates may include reminders to leave early for meetings due to traffic conditions or suggestions to switch from coffee to water after multiple caffeine orders during the day.

The update will also add ride-hailing, flash shopping, itinerary planning, and movie ticket booking features later this month. These additions are designed to reduce the need to frequently access smartphones during commuting, travel, or leisure activities.

Another major addition is what Alibaba describes as the world’s first spatial 3D display system for AI glasses. The Qwen AI Glasses S1 uses dual optical engines and binocular stereoscopic imaging technology to create layered visuals with depth perception instead of flat overlays.

This could make navigation prompts, notifications, subtitles, and digital content appear more naturally within the user’s field of view. Alibaba also announced partnerships related to tourism and cultural projects, including AI-assisted documentary creation and travel-focused experiences.

The update is rolling out to the Qwen AI Glasses S1, which is also marketed as the Quark AI Glasses S1 in China.

For more daily updates, please visit our News Section.

Stay ahead in tech! Join our Telegram community and sign up for our daily newsletter of top stories! 

The post Alibaba’s Qwen AI Glasses S1 gets proactive AI features and spatial 3D display upgrade appeared first on Gizmochina.

The consumer PC hardware market is reportedly heading toward one of its biggest downturns in years. According to a new report from DIGITIMES, major Taiwanese motherboard brands have sharply reduced their shipment targets for 2026 as rising hardware prices and weak demand continue to hurt the industry. 

Companies including ASUS, Gigabyte, MSI, and ASRock are all expected to see major shipment declines next year.

AI Demand Is Hurting Consumer Hardware Supply

One of the biggest reasons behind the slowdown is the rapid expansion of the AI industry. Chipmakers like Intel and AMD are reportedly prioritizing production of high-profit server processors such as Xeon and EPYC chips instead of consumer CPUs. As a result, desktop and laptop processors are facing supply shortages and price increases. Reports also suggest that additional CPU price hikes may arrive later in 2026.

RAM and Storage Prices Continue to Rise

Memory prices have also increased sharply over the past year. Reports claim RAM and storage now account for over 30% of a PC’s total build cost, compared to around 15% previously. Because of these rising costs, many laptop makers have increased prices by 10–20%, while some affordable models are now offering weaker specifications to keep prices under control. This has reduced consumer purchasing power and slowed upgrade demand.

GPU Market Shows Little Excitement

The gaming GPU market is also struggling to attract buyers. NVIDIA has not introduced major RTX 50 series upgrades in 2026, while rumors suggest the RTX 60 series may not launch until 2028. GPU prices also remain high in many regions, making gaming PC upgrades less attractive for consumers. Many users are reportedly choosing to keep their older PCs longer instead of spending heavily on new hardware.

Motherboard Brands Face Shipment Declines

ASUS is reportedly struggling to maintain even 10 million motherboard shipments in 2026 after shipping around 15 million units in 2025. Gigabyte, MSI, and ASRock are also expected to see major declines in motherboard sales as the overall DIY PC market weakens further.

AI Servers Become the Main Revenue Driver

Despite weaker consumer PC sales, companies are still expected to remain profitable thanks to booming AI server demand. AI servers are now generating significantly more revenue than traditional consumer hardware, helping brands offset declining motherboard and graphics card shipments.

Read More:

• Samsung Galaxy Watch Can Predict Fainting Spells, Clinical Study Reveals
• Samsung Turned Smartphone Screens Into Health Monitors: Blood Pressure, Heart Rate and More
• Honor 600 China Variant Could Pack Up to 9000mAh Battery, Snapdragon 8 Elite

(via)

The post Taiwanese Motherboard Market Faces Major Collapse in 2026 appeared first on Gizmochina.

Samsung has announced a major advancement in wearable healthcare technology. A new clinical study conducted in partnership with Chung-Ang University Gwangmyeong Hospital in South Korea found that the Samsung Galaxy Watch6 can predict fainting episodes before they happen.

The research focused on vasovagal syncope (VVS), one of the most common causes of fainting. According to the study, the smartwatch was able to warn users up to five minutes before a fainting episode with an overall accuracy of 84.6%.

What is Vasovagal Syncope?

Vasovagal syncope occurs when a person experiences a sudden drop in heart rate and blood pressure. This can temporarily reduce blood flow to the brain, causing loss of consciousness.

Common triggers include stress, fear, emotional shock, anxiety, and pain. While fainting itself is usually not life-threatening, sudden falls can lead to serious injuries such as fractures, head trauma, and concussions.

Researchers noted that up to 40% of people may experience VVS at least once during their lifetime, while many patients suffer repeated episodes.

How the Galaxy Watch6 Detected Fainting Risk

The study involved 132 patients with suspected VVS symptoms. Under medical supervision, participants underwent induced fainting tests while wearing the Galaxy Watch6.

The smartwatch used its built-in photoplethysmography (PPG) sensor to collect heart rate variability (HRV) data. Researchers then analyzed the data using an AI-based prediction system.

The results showed:

• 84.6% overall prediction accuracy
• 90% sensitivity
• 64% specificity
• Up to 5 minutes of advance warning

Samsung says this is the world’s first successful demonstration of a commercially available smartwatch predicting fainting spells in advance.

Samsung Plans More Preventive Health Features

Professor Junhwan Cho, who led the research, said early warnings could give patients enough time to sit down, lie safely, or seek help before collapsing.

Samsung also stated that it plans to expand AI-powered preventive healthcare features in future Galaxy wearables. The findings of the study were published in the European Heart Journal, Digital Health.

Read More:

• Qualcomm Snapdragon 6 Gen 5 and 4 Gen 5 Announced With Wi-Fi 7, 90FPS Gaming
• Samsung Turned Smartphone Screens Into Health Monitors: Blood Pressure, Heart Rate and More
• XPPen Magic Note Pad Feels Like Real Paper – Price, Specs & Discount

(via)

The post Samsung Galaxy Watch Can Predict Fainting Spells, Clinical Study Reveals appeared first on Gizmochina.

XPPen Magic Note Pad has officially launched in India, perfect for students, professionals, and creators who want a paper-like digital writing experience. With a focus on note-taking, reading, and light productivity, this device blends the feel of traditional notebooks with modern Android features. It aims to stand out in a growing segment of smart writing tablets by offering eye comfort, precision stylus input, and built-in productivity tools.

Display and Writing Experience

The Magic Note Pad features a 10.95-inch X-Paper display designed to mimic real paper. The AG nano-etched surface reduces glare and reflections, making it comfortable for long reading or writing sessions. It supports 16.7 million colors, 400 nits brightness, and a smooth refresh rate for clear visuals.

A dedicated X-Paper Key lets users switch between three modes: Ink, Light, and Natural, depending on the task. Whether you’re sketching, taking notes, or reading documents, the screen adapts easily, making it versatile for everyday use.

Performance and Features

Running on Android 14, the device supports apps from the Google Play Store, giving it more flexibility than traditional e-note devices. It comes with 128GB of storage, which is enough for notes, PDFs, and apps.

The XPPen Notes app is built-in and offers features like handwriting-to-text conversion, PDF annotation, auto-save, and cloud syncing with Google Drive and OneDrive. The included X3 Pro Pencil 2 provides 16,384 pressure levels, ensuring accurate and smooth writing. It also supports magnetic attachment and shortcut keys for better productivity.

For remote work, the tablet includes a 13MP front camera, dual speakers, and microphones, making it suitable for video calls and online meetings.

Battery and Design

The Magic Note Pad packs an 8000mAh battery with 20W fast charging, designed to last through a full day of usage. It is slim at 7mm and lightweight at 495g, making it easy to carry around.

The display is TÜV SÜD certified for low blue light, which helps reduce eye strain during long usage sessions.

Price and Discount Offer

XPPen Magic Note Pad is priced at Rs. 39,999, but it is currently available at a discounted price of Rs. 38,999. Buyers can get an additional 5% discount using the code GizmoXP, making it a more attractive deal for early adopters. Overall, the device positions itself as a strong option for users looking for a paper-like digital writing experience with modern smart features.

The post XPPen Magic Note Pad Feels Like Real Paper – Price, Specs & Discount appeared first on Gizmochina.

OpenAI CEO Sam Altman has pushed back against growing fears that artificial intelligence will replace human jobs, calling such concerns “overly pessimistic” in the long run. Speaking on X (formerly Twitter), Altman emphasized that OpenAI’s goal is to build tools that “augment and elevate” human capabilities rather than replace people entirely.

His comments come at a time when layoffs across industries, especially in the United States, have raised alarm. Several companies have linked job cuts directly to AI adoption, fueling concerns that traditional roles could become obsolete. A widely discussed example is King, the company behind Candy Crush Saga, which reportedly laid off developers after they created AI tools capable of generating game levels.

The anxiety has been further intensified by statements from Dario Amodei, CEO of Anthropic. Amodei recently suggested that AI could soon write nearly all code within the next 6 to 12 months. Such predictions have strengthened the belief that AI may dominate large parts of the workforce.

Altman, however, disagrees with this outlook. He argues that while AI will transform jobs, it will not eliminate the need for humans. Instead, he believes people will shift toward more meaningful and higher-value work. According to him, adaptation, not replacement, will define the future of employment.

Looking ahead, Altman envisions a world where AI reduces the need for intense labor, allowing people to live more flexible and fulfilling lives. While debate continues within the industry, his stance highlights a more optimistic path where humans and AI work together rather than compete.

Read More:

• Huawei Pura 90 Pro vs Xiaomi 17 Pro Full Comparison: Specs, Camera, Battery, Price & Verdict
• Vivo X300 Ultra Global Now on Giztop: 200MP Cameras, 6600mAh Battery & 100W Charging,
• OPPO Find X9 Ultra vs iPhone 17 Pro Max: The Ultimate Flagship Battle

(via)

The post Will AI Take Your Job? OpenAI CEO Sam Altman Gives a Surprising Answer appeared first on Gizmochina.

Anker has introduced its first in-house AI chip platform called “Thus.” The new chip will officially debut on May 21, 2026, during its launch event in New York City. This marks the company’s move into custom silicon, focusing on bringing advanced AI features directly to compact devices like wireless earbuds.

Built for Speed and Efficiency

The Thus chip is based on Compute-in-Memory (CIM) architecture, which combines processing and memory into a single unit. Instead of moving data between separate components, it processes data where it is stored. This reduces latency, improves speed, and lowers power consumption, key for small, battery-powered devices. Anker claims the chip delivers up to 150× more AI computing power compared to its earlier audio products.

Smarter Audio with On-Device AI

A major highlight of the Thus chip is real-time AI processing directly on the device. This allows features like noise cancellation and voice enhancement to work faster without relying on cloud processing. It also improves privacy and ensures more stable performance in different environments.

First Product: AI-Powered Earbuds

The chip will first appear in a new flagship pair of earbuds launching alongside it. These earbuds are expected to feature advanced “Clear Calls” technology using AI-based environmental noise cancellation. The setup includes eight microphones and bone-conduction sensors to filter background noise and improve call clarity.

Anker plans to expand the Thus chip beyond audio products into mobile accessories and other smart devices. The goal is to bring faster, more efficient on-device AI to a wider range of everyday products.

Read More:

• Huawei Pura 90 Pro vs Xiaomi 17 Pro Full Comparison: Specs, Camera, Battery, Price & Verdict
• OPPO Find X9 Ultra Hasselblad Earth Explorer Master Set, Now Available at Giztop
• OPPO Pad 5 Pro Launched: 13.2” 3.4K Display, Snapdragon 8 Elite Gen 5 , Now Available On Giztop

(via)

The post Anker’s New “Thus” Chip Brings 150x AI Power to Earbuds – Launching May 21 appeared first on Gizmochina.

OpenAI might be looking beyond just software. A new report suggests the company is exploring an “AI agent phone,” and the idea behind it is quite different from how smartphones work today.

The details come from analyst Ming-Chi Kuo, who claims OpenAI is thinking about a device where you don’t really move between apps the way you normally would. Instead, a built-in AI agent handles tasks for you, end to end.

So rather than opening your calendar, checking messages, and switching between apps to plan something, you would just ask. The AI would figure out the rest, pulling in the right information, suggesting options, and even completing actions if needed. It sounds simple on paper, but it is a big shift from the current app-based model.

We have seen early attempts at this before. The Rabbit R1 tried to reduce reliance on apps, though in a much more limited way. What OpenAI is reportedly working on seems more ambitious, basically bringing that idea into a full smartphone rather than a companion gadget.

On the hardware side, the report mentions collaborations with MediaTek and Qualcomm to build a chip that is better suited for AI-heavy tasks. Manufacturing could be handled by Luxshare, which already works with large-scale consumer electronics.

That said, this is not something arriving anytime soon. Kuo suggests specs and suppliers could be finalized around late 2026 or early 2027, with production possibly starting in 2028. So even in the best case, this is still a few years out.

There are also some obvious hurdles. Replacing apps, or even just reducing their role, means dealing with a lot of existing ecosystems and habits. People are used to how phones work, and changing that is not easy.

Still, if this direction holds, it gives a glimpse of where things might be heading. Less tapping through apps, more just asking for what you want and letting the system handle it.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source)

The post OpenAI’s first smartphone aims to replace apps with AI agent appeared first on Gizmochina.

An AI coding agent has reportedly deleted an entire production database along with its backups, turning what should have been a routine maintenance session into a serious incident for a startup.

The issue came to light after PocketOS founder Jer Crane shared what happened on April 24. While working in a test environment, he ran into a mismatch with account credentials. Instead of waiting for input or asking for clarification, the AI agent decided to act on its own.

The agent, running through Cursor and powered by Anthropic’s Claude Opus 4.6 model, searched through the codebase, found an API token in a separate file, and used it to execute a GraphQL command. That command ended up deleting a storage volume tied to the company’s production database. The entire sequence reportedly took just a few seconds.

What made things worse is how the backups were set up. They were stored on the same volume, so when the main data was deleted, the backups went with it. The most recent usable backup was around three months old, which meant a lot of recent data was at risk of being lost.

It later turned out that the API token used by the agent had far more access than intended. It was originally created for something limited, like managing domains, but effectively had root-level permissions due to missing role-based restrictions. That gave the agent the ability to carry out a destructive command without hitting any safeguards.

When asked to explain what happened, the AI reportedly acknowledged that it made assumptions about the environment, did not verify what the command would do, and went ahead without proper authorization.

The incident quickly picked up attention online, with posts about it spreading widely. Railway’s CEO, Jake Cooper, stepped in to help with recovery, and the system was brought back online within about an hour. Since then, a delayed deletion mechanism has been introduced to avoid instant data wipes.

Even so, the situation highlights something that has been building for a while. As AI tools become more capable, giving them direct access to production systems without strict limits can backfire quickly. In this case, it only took 9 seconds.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source: @lifeof_jer | Image)

The post AI coding agent wipes out entire production database in 9 seconds appeared first on Gizmochina.

CHUWI AuBox X is a new ultra-compact mini PC designed for the AI-powered computing era. With a tiny 0.67L palm-sized body, it focuses on delivering strong performance without taking up desk space. It comes pre-installed with Windows 11 Pro and is built for creative professionals, home offices, and modern work setups. Despite its size, it aims to handle demanding tasks like content creation, multitasking, and AI workflows efficiently.

AI Performance and Capabilities

One of the biggest highlights of the AuBox X is its AI power. It delivers up to 115 TOPS on the Ultra 7 model and 97 TOPS on the Ultra 5. This performance comes from a combination of CPU, GPU, and a dedicated NPU. The system is designed to run AI tasks locally, meaning no cloud dependency. This results in faster speeds, lower latency, and better privacy. Tasks like photo enhancement, real-time background removal, and AI-assisted workflows run smoothly on-device.

Dedicated AI Hardware and Software

The mini PC includes an integrated Intel AI Boost NPU, offering up to 47 TOPS on the Ultra 7. This enables efficient AI processing and supports advanced features like live captions, smart search improvements, and creative tools. It also supports an open-source AI assistant, which can run on Windows or through WSL. This setup allows continuous AI operations with low power usage, making it ideal for long sessions.

OpenClaw AI Assistant and Automation

CHUWI AuBox X also supports OpenClaw, an open-source AI assistant designed for automation and continuous workflows. It can run directly on Windows or through WSL (Windows Subsystem for Linux), giving users flexibility depending on their setup. Since it runs locally, OpenClaw delivers near zero-latency responses while keeping all data private. It can handle tasks like automation, AI assistance, and background processing 24/7 with low power usage. This makes the system especially useful for developers, creators, and users who want full control over their AI tools without relying on cloud services.

Performance and Graphics

Powered by Intel’s Lunar Lake processors, the AuBox X comes in two variants: Ultra 5 and Ultra 7. Both feature 8 cores and boost speeds up to 4.8GHz with configurable power usage. The integrated Intel Arc graphics deliver impressive results for a mini PC. The higher model offers performance close to entry-level dedicated GPUs, making it capable of handling photo editing, light video work, and even some 3D tasks. Despite this, the system remains quiet, with noise levels under 45dB.

Connectivity and Expansion

The AuBox X offers strong connectivity options, including USB4 with eGPU support, HDMI 2.1, and multiple USB-A ports. It supports up to three displays and can output up to 8K resolution. Networking includes 2.5Gb Ethernet, Wi-Fi 6E, and Bluetooth 5.3. Storage and memory are also fast, with LPDDR5X RAM and dual SSD slots supporting high-speed PCIe drives.

Performance Variants 

The Ultra 5 model is ideal for everyday tasks, office work, and media consumption. It offers balanced performance at a more affordable price. The Ultra 7 model is better suited for creative professionals who need more power for video editing, AI tasks, and heavier workloads.

Pricing and Availability

Pre-orders begin in mid-April, with pricing starting at $699 for the Ultra 5 and $829 for the Ultra 7. An early-bird discount of $70 is also available for a limited time, making it a competitive option in the compact PC market.

CHUWI AuBox X stands out by combining a compact design with serious AI capabilities. It offers a strong balance of performance, efficiency, and modern features, making it a smart choice for users looking to upgrade to an AI-ready desktop without sacrificing space.

The post CHUWI AuBox X Launched: Tiny AI Mini PC with 115 TOPS Power appeared first on Gizmochina.

Nvidia CEO Jensen Huang has issued a strong warning about the future of artificial intelligence. He said that if DeepSeek’s upcoming AI model runs entirely on Huawei chips, it could be a “horrible outcome” for the United States. His concern is that global AI systems may start shifting away from US technology, which could weaken America’s leadership in this space.

DeepSeek V4 Could Change the Game

The upcoming DeepSeek V4 model, expected later this month, is already creating buzz. Reports suggest it could run on Huawei’s Ascend chips, marking a major shift from the current reliance on Nvidia GPUs. However, some signs indicate it may still support Nvidia’s latest Blackwell architecture. If DeepSeek fully moves to Huawei hardware, it would be the first major AI model to break away from the US ecosystem.

China’s Different Approach to AI

Huawei’s chips are still considered slightly behind US alternatives in raw power. But China is focusing on a different strategy. Instead of relying on the most advanced chips, companies are using more chips together and improving software efficiency. Techniques like Mixture-of-Experts help boost performance without needing the best hardware.

Strength in Scale and Resources

China has key advantages that support this approach. It has a large pool of AI talent and access to cheaper energy. This allows companies to run massive computing setups. Even with weaker chips, using many processors at once can deliver strong results.

Rising Tensions and Policy Pressure

The situation is becoming more complex due to growing US–China tensions. Lawmakers in the US are pushing for stricter restrictions on Chinese AI firms. At the same time, some chip supply decisions suggest partial easing, showing mixed signals from policymakers.

This is no longer just about hardware power. The real battle is shifting toward software, efficiency, and ecosystem control. If DeepSeek succeeds with Huawei chips, it could mark a turning point in global AI, potentially reshaping who leads the industry in the coming years.

Read More:

• DJI Lito X1 Series Launching April 23: Expected Specs, Price, and Features
• Lenovo May 19 Event Confirmed: Legion Y70, Y900 Tablet, Razr Fold Incoming
• Galaxy Z TriFold Sold Out Forever – Here’s Why

(via)

The post Nvidia CEO Issues Stark Warning: Huawei-Powered AI Could Shake US Dominance appeared first on Gizmochina.

Japan has approved major financial support for Sony Group to strengthen domestic semiconductor production, with a clear focus on AI image sensors.

The subsidy, worth up to 60 billion yen (around $380 million), has been granted under Japan’s Economic Security Promotion Law by the Ministry of Economy, Trade and Industry. The move is part of a broader national effort to secure a stable supply of critical semiconductor components and reduce dependence on external manufacturing.

Sony Semiconductor Solutions plans to invest a total of 180 billion yen (around $1.13 billion) to build a new semiconductor facility in Koshi, Kumamoto Prefecture. Construction is already underway. The plant is expected to produce 10,000 units per month of 300mm wafers, with supply scheduled to begin in May 2029.

The government has identified image sensors as essential for future technologies. Key applications include autonomous vehicles, automated industrial systems, and AI-driven imaging solutions. Economy Minister Ryosei Akazawa highlighted the importance of ensuring a stable supply of these components for the growth of physical AI.

Sony already leads the global image sensor market, with its sensors widely used in smartphones, automotive systems, and other advanced technologies.

The development reflects Japan’s broader push to strengthen its semiconductor ecosystem and stay competitive in AI and next-generation imaging technologies.

Read More:

• Leak Suggests Redmi Could Launch Xiaomi’s First 10,000mAh Battery Phone
• Huawei Watch GT 6 Pro Gets HarmonyOS 6.0 Update with Nap Tracking, New Watch Faces and More
• Honor April 23 Launch: New MagicBook X Series, Gaming Laptops, and More

(via)

The post Japan Pumps $380M Into Sony to Secure Future of AI and Autonomous Tech appeared first on Gizmochina.

Highlights:

• Features one of the most advanced AI projection headlight systems ever
• Capable of displaying real-time road signals and alerts
• Up to 750 km range (BEV) with dual-motor AWD standard
• Level 4-ready tech with steer-by-wire system

XPeng is set to officially launch its new flagship SUV, the GX, on April 15, bringing a strong focus on intelligent lighting and advanced driving technology. Developed in collaboration with Huawei Qiankun, the GX introduces a new generation of AI-powered digital projection headlights designed to transform how vehicles interact with their surroundings.

The standout feature is its advanced DLP projection lighting system. Unlike traditional headlights, the GX can project dynamic symbols directly onto the road. This includes arrows for lane guidance, “Yield to Pedestrians” alerts, and even following distance indicators. The system also supports SOS projections and personalized welcome animations, adding both safety and premium appeal. It can even project videos, turning the vehicle into a mobile outdoor cinema.

In terms of performance, the GX will be available in both all-electric (BEV) and extended-range (EREV) versions. The BEV model offers a claimed range of up to 750 km (CLTC), while the EREV delivers 320 km of pure electric driving. All variants come with dual-motor all-wheel drive and a top speed of 200 km/h.

The SUV features a large, premium six-seater layout with a length of 5265 mm and a 3115 mm wheelbase. Design highlights include a yacht-inspired body, clamshell hood, split tailgate, and a panoramic glass roof. Inside and out, the GX focuses on blending minimalism with a bold, futuristic identity.

With Level 4 autonomous driving capability support and a next-gen steer-by-wire system, the XPeng GX positions itself as a technology-first flagship in the large SUV segment.

Read More:

• Alibaba’s Qwen Dominates Open-Source AI: Nears 1 Billion Downloads
• China’s Unitree Sells Cheapest Humanoid Robots Online with R1 Global Launch
• Xiaomi Robot Vacuum and Mop 6 debuts with 28,000 Pa suction, 80°C hot water cleaning, and obstacle avoidance

(via)

The post XPeng with Huawei Launches the World’s Most Advanced AI Headlights Yet appeared first on Gizmochina.

Highlights:

• Apple is developing its own AI server chip, codenamed “Baltra,” for cloud infrastructure
• Built using advanced 3nm technology from TSMC
• Move aims to reduce reliance on NVIDIA and strengthen Apple’s AI ecosystem

Apple is quietly building a major foundation for its AI future with a new self-designed server chip known internally as “Baltra.” This chip is expected to power the company’s internal cloud infrastructure, focusing on secure data processing and backend AI workloads. It marks a major shift as Apple expands beyond consumer hardware into large-scale AI computing.

The Baltra chip is likely to be manufactured by TSMC using its second-generation 3nm (N3E) process. This technology is designed to deliver improved performance and efficiency compared to earlier nodes. Apple is also investing heavily in TSMC’s advanced SoIC (System on Integrated Chips) packaging, which enables vertical stacking of multiple chip components for better speed and power efficiency.

Reports suggest Baltra will use a chiplet-based architecture, allowing different specialized chips to work together in one package. This design improves scalability and lets Apple optimize AI workloads more effectively. The company is also working with partners on interconnect technologies while gradually aiming to bring more design capabilities in-house.

To support its AI push, Apple has reportedly reserved significant production capacity at TSMC for the coming years. A large portion of this is expected to be allocated to AI server chips like Baltra, signaling serious long-term investment.

Overall, this move highlights Apple’s plan to control its entire AI stack, from silicon and packaging to cloud infrastructure, while reducing dependence on external chip providers and competing more directly in the AI hardware space.

Read More:

• Alibaba’s Qwen Dominates Open-Source AI: Nears 1 Billion Downloads
• China’s Unitree Sells Cheapest Humanoid Robots Online with R1 Global Launch
• Samsung Expanding AI Anti-Fraud Feature Beyond Galaxy S26, Foldables Next

(via)

The post Apple May Team Up with TSMC for Secret AI Chip ‘Baltra’ appeared first on Gizmochina.

Highlights:

• Samsung’s AI anti-fraud feature can detect scam calls in real time
• Powered by on-device Gemini AI, originally from Google Pixel phones
• Global rollout expected, with foldables like Z Fold 8 and Flip 8 likely to get it

Samsung is preparing to expand its new AI-powered anti-fraud feature beyond the Galaxy S26 series, signaling a wider global rollout in the coming months. The feature, currently limited to users in the United States, is designed to detect scam calls in real time and warn users instantly if a conversation appears suspicious.

The technology behind this feature comes from Google and uses an on-device Gemini AI model. This means call content is analyzed directly on the phone, ensuring better privacy since data does not need to be sent to external servers. The system listens for patterns commonly linked to fraud and alerts users during the call itself.

Samsung has integrated this capability into its native Phone app through a collaboration with Google. Interestingly, this AI feature was first introduced on Pixel smartphones before making its way to Samsung’s flagship lineup.

Recent findings from code strings in the Google Phone app (version 217 beta) suggest that Samsung is planning to bring this feature to more devices. These include upcoming foldables such as the Galaxy Z Fold 8 and Galaxy Z Flip 8, along with other regional variants across global markets.

The presence of multiple international model numbers strongly indicates that Samsung is aiming for a large-scale rollout. However, one notable absence is the expected global variant of the Flip 8, leaving some uncertainty.

Overall, this move suggests that AI-powered fraud detection could soon become a standard feature across Samsung Galaxy devices worldwide.

Read More:

• Alibaba’s Qwen Dominates Open-Source AI: Nears 1 Billion Downloads
• China’s Unitree Sells Cheapest Humanoid Robots Online with R1 Global Launch
• Keychron K3 Max SE Goes Global at $104.99 – Ultra-Slim QMK Keyboard You Shouldn’t Miss

(via)

The post Samsung Expanding AI Anti-Fraud Feature Beyond Galaxy S26, Foldables Next appeared first on Gizmochina.

Unitree Robotics is set to launch its cheapest humanoid robot, the R1, in global markets next week. What makes this launch stand out is the sales strategy. Unitree is taking humanoid robots directly to online platforms, starting with AliExpress. This marks a shift from traditional enterprise sales to a more accessible, e-commerce-driven approach.

The initial rollout will cover North America, Europe, Japan, and Singapore. The company is also joining Alibaba’s Brand+ channel, which brings benefits like free shipping, free returns, and better global visibility. More online and offline sales channels are expected to follow.

R1 Robot: Built for Motion, Priced for Access

The R1 was first launched in China in 2025 at a starting price of 29,900 yuan (around $4,370), making it one of the most affordable humanoid robots available. While global pricing is still unannounced, the company is expected to keep it competitive.

Standing 123 cm tall and weighing over 27 kg, the R1 is designed for dynamic movement. It can perform cartwheels, run downhill, stand up from the ground, and execute complex routines. The “Born for sport” positioning highlights its focus on agility and motion rather than industrial tasks.

From Labs to Online Buyers

Unitree’s move to sell robots online signals a major shift in how humanoid machines are marketed and distributed. Until now, most robots were sold through enterprise deals or research contracts. By listing on platforms like AliExpress, Unitree is opening access to developers, educators, and even early consumers globally.

This builds on its existing customer base, where around 70% of shipments in 2025 went to universities and research institutions. The online model could further expand this ecosystem.

Scaling Fast with a Low-Cost Strategy

The company shipped over 5,500 humanoid robots in 2025, far ahead of competitors like Tesla, Figure AI, and Agility Robotics, which delivered around 150 units each. Unitree now aims to ship 10,000 to 20,000 units in 2026.

Its cost advantage comes from a highly localized supply chain, with over 80% of components sourced within China. This allows it to price robots far below global averages, where similar machines can cost up to $300,000.

A New Phase for Consumer Robotics

Unitree’s online-first global launch could reshape the humanoid robotics market. By combining low pricing with e-commerce accessibility, the company is testing real-world demand beyond labs and factories.

If successful, this approach could accelerate adoption and bring humanoid robots closer to mainstream buyers much faster than expected.

Read More:

• Alibaba’s Qwen Dominates Open-Source AI: Nears 1 Billion Downloads
• Xiaomi Mijia Robot Vacuum Mop 4 launches with 10000Pa suction, LDS system, long battery life
• Humanoid Robots Are Scaling Fast, And AGIBOT from China Is Leading the Charge

The post China’s Unitree Sells Cheapest Humanoid Robots Online with R1 Global Launch appeared first on Gizmochina.

Highlights:

• Living rat neurons successfully trained to perform real-time AI computations
• Structured neuron networks improved learning and reduced synchronization
• Technology shows strong potential for brain-machine interfaces and bio-AI systems

Research Overview

Researchers developed a novel bio-AI system where living rat cortical neurons were trained to perform real-time computational tasks. The study focuses on combining biological neural networks with machine learning techniques using a closed-loop reservoir computing approach. The goal is to explore whether living neurons can act as functional computing systems rather than just biological components.

System Design and Working

The system integrates living neurons with high-density microelectrode arrays and microfluidic devices. Neural signals are recorded, converted into continuous outputs, and fed back into the system as electrical stimulation in a loop of about 330 milliseconds. A real-time learning method continuously adjusts the output to match target signals, allowing the system to learn without external intervention.

Network Structuring Innovation

To improve performance, neurons were physically organized into 128 micropores connected through microchannels. This design prevented excessive synchronization, which is common in unstructured neural networks. As a result, neuron correlation dropped significantly from 0.45 to around 0.12, leading to more complex and efficient network behavior. The lattice network structure showed the best overall performance.

Capabilities and Results

The system successfully generated different waveform patterns such as sine, square, and triangular waves across multiple time intervals. It also demonstrated the ability to approximate complex chaotic systems like the Lorenz attractor. During training, the system maintained strong accuracy, achieving correlation levels above 0.8.

Limitations and Future Scope

Despite its capabilities, the system shows a performance decline after training stops, with increasing error during autonomous operation. A key limitation is the 330-millisecond feedback delay, which restricts the system’s ability to handle fast-changing signals. Future work aims to reduce latency using specialized hardware, with potential applications in brain-machine interfaces, neural prosthetics, and next-generation bio-hybrid AI systems.

Read More:

• Xiaomi’s New Electric Shaver Pro Set is Insane – 95 Days Battery & Smart Cleaning
• Inside Project Prometheus: The Secret AI Startup of Jeff Bezos, That’s Hiring Top xAI Talent
• Huawei Unveils Vision Smart Screen 6 SE: 288Hz 4K Mini LED TV Goes Official

(via)

The post Rat Brain Cells Trained to Perform AI Tasks in Groundbreaking Study appeared first on Gizmochina.

• Backed by Jeff Bezos, Prometheus aims to build real-world industrial AI
• Former xAI co-founder Kyle Kosic joins to lead infrastructure
• Plans to raise billions and invest in AI-disrupted industries

A new and highly secretive AI startup, Project Prometheus, is quietly emerging as one of the most ambitious bets in the artificial intelligence race. Backed by Amazon founder Jeff Bezos and co-led by former Google executive Vikram Bajaj, the company was founded in 2025 with a bold goal: to take AI beyond chatbots and into the physical world.

Unlike tools such as ChatGPT or Claude Code, Prometheus is focused on building systems that understand real-world environments. Its AI models are being designed to work in industries like aviation, engineering, and architecture, using domain-specific data such as jet engine design.

A major boost to the company comes from the hiring of Kyle Kosic, a co-founder of xAI. Kosic, who previously worked with Elon Musk and helped build the Colossus supercomputer, will now lead Prometheus’ AI infrastructure efforts.

The startup is also part of a growing “talent war” in AI, where top engineers are being aggressively recruited with massive offers. Notably, all 11 co-founders of xAI have now exited, highlighting rising tensions in the industry.

Prometheus is expanding fast, with offices in San Francisco, London, and Zurich, and hundreds of employees already hired. Its long-term plan includes raising tens of billions of dollars and building a Berkshire Hathaway-style investment model, acquiring stakes in companies, and using their data to train smarter AI.

Despite its rapid growth, Prometheus remains highly secretive, with no official public statements yet.

Read More:

• Xiaomi’s New Electric Shaver Pro Set is Insane – 95 Days Battery & Smart Cleaning
• Xiaomi’s new robot vacuum and mop packs 10,000Pa suction, 180-min runtime at an affordable price
• Humanoid Robots Are Scaling Fast, And AGIBOT from China Is Leading the Charge

(via)

The post Inside Project Prometheus: The Secret AI Startup of Jeff Bezos, That’s Hiring Top xAI Talent appeared first on Gizmochina.

NVIDIA’s DLSS 5 rollout hasn’t exactly been smooth, and now there’s another twist. The company’s own promotional trailer has disappeared from YouTube, but not for the reason you’d expect.

The video, which was first shown at GTC last month, highlighted DLSS 5’s new neural rendering capabilities. The demo clips, including scenes from Resident Evil: Requiem, drew attention for their lighting and overall visual quality. At the same time, though, the tech quickly became a talking point for less positive reasons.

A lot of the criticism has focused on how DLSS 5 affects visuals, especially character faces. Some players feel the AI processing smooths things out too much, giving scenes a slightly artificial look. The reaction online has been fairly divided, and in some cases quite harsh. Reports suggested the official trailer had a low like ratio despite pulling in millions of views.

NVIDIA CEO Jensen Huang has addressed those concerns more than once, saying developers still have control over how the tech is used and that it’s meant to support, not override, a game’s original style. Still, the debate hasn’t really settled.

Now the situation has taken a more unusual turn. The DLSS 5 trailer on NVIDIA’s GeForce YouTube channel was removed, but it wasn’t taken down by NVIDIA itself. Instead, it was pulled following a copyright claim.

According to reports, Italian broadcaster La7 used parts of the trailer in a program and then ended up triggering a copyright claim through YouTube’s system. The result was that not only NVIDIA’s original video, reportedly with over 2 million views, but also other uploads using the same footage were blocked.

It’s the kind of issue that points back to YouTube’s automated Content ID system, which doesn’t always get things right. In this case, the trailer is clearly NVIDIA’s own content, so the takedown is likely temporary. Still, it adds another layer to what has already been a somewhat messy rollout for DLSS 5.

For now, the conversation around the technology, and how it’s being used, doesn’t seem to be slowing down anytime soon.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

The post YouTube took down NVIDIA’s own DLSS 5 trailer following a copyright issue appeared first on Gizmochina.

AGIBOT has announced a major achievement, rolling out its 10,000th humanoid robot on March 30, 2026, in Shanghai. This makes it one of the first robotics companies in the world to reach such a scale. The milestone reflects a clear shift in the industry, from early-stage testing to large-scale real-world deployment. It also signals that humanoid robots are no longer limited to labs or experiments but are becoming part of everyday business environments.

From Experiment to Real-World Use

This achievement is not just about production numbers. It highlights AGIBOT’s ability to deploy robots at scale across different industries. The company is moving beyond niche use cases and entering mainstream commercial adoption. This marks an important transition where the focus is no longer just on whether the technology works, but on how it delivers real business value. It also shows that embodied AI is becoming practical and usable in real-world settings.

Rapid Production Growth

AGIBOT’s growth has accelerated significantly over time. It took nearly two years to produce its first 1,000 robots. The jump from 1,000 to 5,000 units took about a year. However, scaling from 5,000 to 10,000 units happened in just three months. This fourfold increase in speed highlights major improvements in manufacturing efficiency and supply chain maturity. It also reflects rising demand for humanoid robots across industries.

According to Peng Zhihui, this rapid scaling is the result of a mature supply chain and standardized manufacturing processes. He emphasized that demand for large-scale commercial robotics solutions is increasing quickly. The company is focused on expanding the adoption of embodied AI systems and making them more accessible to businesses worldwide.

Expanding Real-World Applications

A large number of AGIBOT’s robots are already deployed in real-world environments. They are being used in logistics, retail navigation, hospitality services, and education. The company is also expanding into industrial applications, including production line support and manufacturing operations. This shows a clear shift toward practical, day-to-day use of humanoid robots.

Global Reach and Industry Impact

AGIBOT is also seeing strong international demand, with deployments across Europe, North America, Japan, South Korea, Southeast Asia, and the Middle East. The focus is moving from pilot projects to repeat large-scale rollouts. This milestone confirms that humanoid robots are now scalable, commercially viable, and ready to be integrated into everyday business operations, positioning AGIBOT as a leader in the global robotics industry.

What Is AGIBOT?

AGIBOT is a Shanghai-based robotics company focused on humanoid robots powered by embodied AI for industrial and service use. Founded in 2023, it has quickly scaled from prototypes to mass production, crossing 10,000 robots and becoming one of the first to deploy humanoids at this level. Its systems are already used in logistics, retail, hospitality, and manufacturing environments. With rapid production growth and expanding global deployments, AGIBOT is now focused on scaling real-world adoption and making general-purpose robots more practical across industries.

Read More:

• China vs US in Robotics: Nvidia CEO Reveals the Real Power Shift
• ANTHBOT M5 Smart Robot Lawn Mower Gets Massive Discount – Limited Time Amazon Deal
• AI-Powered Robots Begin Real Battlefield Testing

The post Humanoid Robots Are Scaling Fast, And AGIBOT from China Is Leading the Charge appeared first on Gizmochina.

OpenAI is reportedly shutting down its Sora video-generation tool, less than two years after it first drew widespread attention. Sora was introduced in 2024 as a way to create realistic video clips from simple text prompts. Early demos generated a lot of buzz, especially around how quickly the quality seemed to improve.

But that momentum didn’t quite translate into long-term traction. OpenAI has now announced that it plans to step back from video generation and focus more on areas like robotics and so-called “agentic” AI.

That said, the situation around Sora isn’t a clean, immediate shutdown. While OpenAI has indicated that it is discontinuing the app and winding down development, the tool hasn’t disappeared overnight. The web version, for instance, is still accessible for now, and there’s no clearly defined cutoff date yet. It suggests Sora is in a transition phase—no longer a priority, but not fully offline either.

The shift isn’t just about the product itself. OpenAI is also walking away from a major content partnership with The Walt Disney Company, which had been announced only a few months ago. The deal, reportedly worth around $1 billion, would have allowed Sora users to generate videos featuring well-known Disney characters.

According to reports, the agreement hadn’t progressed far enough for any payments to be made, and Disney was caught somewhat off guard by the decision. The company has since said it will continue exploring AI partnerships elsewhere, while keeping a close eye on how its intellectual property is used.

Sora’s commercial performance may have played a role here. Despite the initial excitement, the platform reportedly brought in around $1.4 million in global in-app revenue—a relatively small figure compared to ChatGPT’s $1.9 billion over a similar period. There were also ongoing concerns around moderation, including misuse involving deepfakes, copyrighted content, and non-consensual imagery.

At the same time, competition in the space has been heating up, with new tools emerging from other companies. Combined with legal risks and rising infrastructure costs, continuing Sora may have been harder to justify.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source)

The post OpenAI discontinuing its AI video generation tool Sora appeared first on Gizmochina.

Key Highlights

• Humanoid robots like Phantom MK-1 are now being tested in real combat environments
• Over 7,000 robot missions completed in a single month show rapid adoption
• Future warfare may rely on coordinated AI robots across land, air, and sea

The use of AI-powered robots in warfare is no longer just an idea; it is becoming a reality. Modern battlefields are now being used to test advanced machines designed to reduce human risk and improve efficiency. These developments show how quickly robotics and artificial intelligence are moving from labs into real-world situations.

Humanoid Robots Enter the Field

One of the most advanced examples is the Phantom MK-1 humanoid robot. It is designed to move like a human and operate in difficult terrains where traditional machines struggle.

The robot stands around 175 cm tall, weighs about 80 kg, and can carry up to 20 kg. It uses cameras and sensors to understand its surroundings and can move at speeds of up to 6 km/h.

These robots are not fully independent. They are being tested to study mobility, performance, and how AI behaves under pressure.

How AI Military Robots Work

Military robots today use a mix of AI and human control. This is called a “human-in-the-loop” system. AI helps with tasks like identifying objects, navigating terrain, and suggesting actions. However, humans still control critical decisions, especially when it comes to using weapons.

Rise of Ground Robots

Humanoid robots are only part of the story. Uncrewed Ground Vehicles (UGVs) are already widely used.

In January 2026 alone, more than 7,000 missions were carried out using robots. These machines mainly handle logistics such as delivering supplies, evacuating injured soldiers, and scouting areas. Most robots are currently used for support tasks rather than direct combat.

Challenges and Future Potential

Despite rapid growth, there are still limitations. Robots face issues like limited battery life, high costs, and difficulty understanding complex situations. There are also concerns about hacking and misuse.

Looking ahead, experts believe future warfare could involve large groups of connected robots working together across land, air, and sea. This shift is not just about warfare; it is a major step forward in robotics and AI. Machines are slowly moving from tools to active partners, shaping the future of technology.

Did You Know? 

The Phantom MK-1 is built by a San Francisco-based startup called Foundation, founded by former military personnel and engineers focused on defense robotics. The company has already secured about $24 million in contracts with the US Army, Navy, and Air Force, making it an official defense partner. 

Beyond this robot, the global race for military robotics is accelerating; countries like the United States, China, Israel, and Russia are actively developing and deploying robotic systems. China has tested armed robot dogs in military drills, while the US has long used systems like PackBot and TALON in combat zones. Even countries like Estonia and Turkey are building advanced unmanned ground and aerial combat systems, showing that the future battlefield is rapidly becoming automated.

Read More:

• US Router Ban Explained: Should You Be Worried?
• Forgot Passwords: Dell’s Latest Mouse Unlocks Your PC
• China’s Military Robot Wolves Revealed: Armed Packs Ready for Urban Warfare

(via)

The post AI-Powered Robots Begin Real Battlefield Testing appeared first on Gizmochina.

• China unveils advanced robotic wolf packs built for urban warfare
• Equipped with micro missiles, grenade launchers, and AI coordination
• Signals rising global race in autonomous military robotics

China has revealed a new generation of robotic “wolf packs,” with first-ever footage showing them operating in simulated urban combat. The development was reported on March 26 by CCTV News, highlighting a major leap in autonomous battlefield technology.

The systems have been developed by China Ordnance Automation Research Institute Co., Ltd. and represent the latest evolution of robotic combat units. Compared to earlier versions, these machines feature a stronger physical build designed to handle harsh environments, along with significantly improved AI “brains” that enable smarter, more autonomous decision-making. Their enhanced coordination also allows them to operate effectively in packs, increasing combat efficiency.

A key highlight is their ability to carry and deploy multiple weapons. These robotic wolves can be equipped with micro missiles and grenade launchers, making them capable of firepower-based target suppression in dense urban warfare scenarios. Their deployment suggests a focus on close-range, high-risk environments where human soldiers face greater danger.

Each unit is assigned a specialized role within the pack. “Shadow” units focus on reconnaissance and scouting, “Bloodstained” units are designed for direct attack, and “Polar” units provide operational support. This structured role system allows for coordinated, team-based tactics similar to human military squads.

Control systems are equally advanced. Operators can guide these machines using voice commands or gesture-based controls via data gloves, enabling real-time human-machine coordination on the battlefield.

Globally, similar technologies are being explored, particularly in the United States. Companies like Boston Dynamics have developed quadruped robots such as Spot, primarily used for surveillance, inspection, and logistics. While the U.S. military has tested armed robotic platforms, it has generally taken a more cautious approach toward deploying fully autonomous combat systems.

The release of this footage signals intensifying competition in military robotics, as nations push toward smarter, more coordinated, and increasingly weaponized autonomous systems.

Read More:

• China vs US in Robotics: Nvidia CEO Reveals the Real Power Shift
• Xiaomi Robot Vacuum 6 Max vs Roborock S9 MaxV Ultra: One Cleans Harder, One Cleans Smarter
• BMW Starts Using Humanoid Robots in Factory to Build Electric Cars

(via)

The post China’s Military Robot Wolves Revealed: Armed Packs Ready for Urban Warfare appeared first on Gizmochina.

TECNO has announced the upcoming beta testing of EllaClaw, a new mobile AI agent built to simplify everyday smartphone tasks. The company positions it as the first OpenClaw-powered AI agent designed specifically for emerging markets. While the exact rollout date is not confirmed, users will soon be able to apply for beta access through official online channels, with more details expected shortly.

What is EllaClaw?

EllaClaw is built using the OpenClaw agentic framework, deeply integrated with TECNO’s own Ella AI assistant. Unlike many AI tools that work as separate apps, EllaClaw operates at the system level, allowing it to interact directly with different parts of the phone.

This setup enables it to handle complex tasks, multi-step workflows, and long-duration processes more efficiently. TECNO is also focusing on making it user-friendly, turning advanced AI technology into something that everyday users can easily understand and use.

Seamless and Easy to Use

One of EllaClaw’s biggest advantages is its simplicity. Users do not need to install or switch between apps. Instead, it works directly within the Ella assistant, offering a smooth and intuitive experience.

The AI is designed to understand natural language, meaning users can give simple or even slightly unclear commands, and the system will still figure out what needs to be done.

Key Features in Beta

EllaClaw introduces several powerful features in its beta version. One of the highlights is one-sentence automation, where users can complete tasks like scheduling, managing files, or performing system actions with just a single instruction.

Another important feature is cross-app data integration. The AI can pull information from apps like SMS, Calendar, and Notes, combining them to provide useful insights. For example, it can summarize messages, highlight important alerts like bank updates or bills, and even suggest deleting unnecessary messages.

It also works as a daily schedule assistant, combining calendar events, notes, weather updates, and news into a simple daily plan that users can follow.

Smarter Over Time with Strong Privacy

EllaClaw includes a persistent memory system, which learns from user behavior over time. This allows it to offer smarter suggestions and become more personalized with regular use.

TECNO has also emphasized privacy and security, stating that user data remains isolated, protected, and not shared with third parties.

Future Plans

Looking ahead, TECNO plans to expand EllaClaw with more localized features tailored for emerging markets. The company also aims to let users build custom AI agents, further boosting productivity.

With its focus on simplicity, automation, and personalization, EllaClaw could become a major step forward in making AI more practical for everyday smartphone users.

Read More:

• Tecno Spark 50 5G teaser shows off a refreshed design
• Tecno Camon 50 Ultra 5G Ranked Best Camera Phone Under $600 by DXOMARK
• TECNO CAMON 50 Ultra 5G Launched With 50MP Sony LYT-700C Camera, Dimensity 7400 and 6500mAh Battery

The post TECNO EllaClaw Beta Announced: A New AI Agent for Emerging Markets appeared first on Gizmochina.

Highlights:

• Huawei launches Atlas 350, focused on AI inference, not training
• Claims up to 2.8× performance boost over Nvidia’s H20 chip
• Powered by Ascend 950PR, part of China’s push for AI independence

New Focus: AI Inference Over Training

Huawei’s Atlas 350 reflects a clear strategic shift toward AI inference, the stage where trained models are deployed in real-world applications. Instead of focusing on training, this accelerator is designed to efficiently run AI systems such as recommendation engines, large language models, and multimodal tools. This aligns with a broader industry transition, where inference is increasingly becoming the main challenge as AI adoption scales across industries.

Performance Designed for Real-World Use

At the heart of the Atlas 350 is Huawei’s Ascend 950PR chip, capable of delivering around 1.56 petaflops of FP4 computing power. Huawei claims this enables up to 2.8 times better performance than Nvidia’s H20 in certain inference workloads. The use of FP4 precision allows faster processing and lower memory usage, making it particularly effective for large-scale deployment where efficiency and speed are more critical than high precision.

Memory and Architectural Advancements

A notable development is Huawei’s move toward in-house high-bandwidth memory, with reported configurations of up to 128GB and 1.6 TB/s bandwidth. This reduces reliance on external suppliers and strengthens China’s semiconductor capabilities. Additionally, the chip is optimized for the “prefill” stage of inference, improving how quickly input data is processed, which is essential for real-time AI applications and emerging agent-based systems.

Building a Complete AI Ecosystem

Huawei’s approach extends beyond a single accelerator. The company is developing an integrated AI ecosystem that includes hardware, storage, and large-scale computing systems. This vertically integrated strategy mirrors established players but is tailored to domestic infrastructure, enabling tighter control over performance and deployment.

Impact on the US–China Tech War

The Atlas 350 is significant for the ongoing US–China technology competition. With restrictions limiting access to advanced US chips, Huawei’s development of competitive domestic alternatives represents a step toward technological self-reliance. For China, this strengthens its ability to build independent AI infrastructure, while for US companies such as Nvidia, it introduces growing competition in a key market.

Implications for Consumers and the Market

For everyday users, this could mean faster and more affordable AI services, especially in China. It may also lead to more local AI apps and platforms. Over time, the tech world could split into separate systems, with different regions building their own AI ecosystems.

Read More:

• China vs US in Robotics: Nvidia CEO Reveals the Real Power Shift
• Xiaomi Robot Vacuum 6 Max vs Roborock S9 MaxV Ultra: One Cleans Harder, One Cleans Smarter
• BMW Starts Using Humanoid Robots in Factory to Build Electric Cars

(via)

The post US – China Tech War: Huawei Takes On Nvidia with Atlas 350 AI Accelerator appeared first on Gizmochina.

Highlights:

• Jensen Huang calls China “formidable” in robotics due to hardware dominance
• Nvidia is betting on Physical AI as the next computing revolution
• Future robotics ecosystem = China’s supply chain + US AI leadership

China’s Robotics Advantage

Jensen Huang highlighted that China currently holds a significant edge in robotics, primarily due to its dominance in core components like microelectronics, motors, rare earth materials, and magnets. These elements are foundational to building robots at scale, and China’s ecosystem is described as the “world’s best.” This deep supply chain strength means even the US robotics industry remains heavily dependent on China for hardware, reinforcing its global influence in the sector.

The Rise of Physical AI

Nvidia is shifting focus beyond traditional GPUs toward what Huang calls Physical AI, the integration of AI into real-world machines like robots and autonomous systems. This marks a major evolution from generative AI to embodied intelligence. Nvidia’s strategy includes building a full-stack ecosystem, from AI models to robotics infrastructure, positioning itself not just as a chipmaker but as the backbone of future intelligent machines.

Nvidia’s Full-Stack Robotics Vision

At its GTC event, Nvidia introduced the Physical AI Data Factory to automate data generation, simulation, and model evaluation. Huang also described a “three-computer” model powering robotics: training systems for AI models, simulation platforms like Omniverse, and edge computers embedded in robots. This approach shows robotics is no longer just hardware; it requires a tightly integrated AI ecosystem.

Market Shifts and Global Competition

While China’s robotics companies, such as Unitree, are scaling rapidly with strong financial growth, Nvidia is navigating a complex market. Its China market share has dropped sharply due to restrictions, but the company is preparing a return with approved H200 AI chips. Despite revenue declines, demand from Chinese firms remains strong.

What Comes Next

Huang believes robotics adoption is just 3–5 years away from widespread use, driven by exponential growth in AI compute and agentic AI systems. The long-term vision points to a massive economic opportunity, where robots augment human labor across industries. The global balance is becoming clear: China dominates the physical layer, while Nvidia aims to control the intelligence powering it.

Read More;

• China’s Top EV Maker BYD Sues U.S. Government: Challenges Trump-Era Tariffs
• BMW Starts Using Humanoid Robots in Factory to Build Electric Cars
• Xiaomi Launches Its Most Advanced Robot Vacuum Yet with Robotic Legs and 35,000Pa Suction

(via)

The post China vs US in Robotics: Nvidia CEO Reveals the Real Power Shift appeared first on Gizmochina.

Highlights:

• BMW begins testing humanoid robots at its Leipzig iFACTORY plant to assist in electric vehicle production.
  
• AEON robot from Hexagon Robotics helps with battery assembly, screw tightening, and other repetitive factory tasks.
  
• AI-powered sensors and self-swapping batteries allow the robot to work continuously with minimal downtime.
  

BMW has begun testing humanoid robots at its iFACTORY plant in Leipzig, Germany, marking an important step toward AI-driven manufacturing. The initiative is part of the company’s plan to integrate advanced robotics and artificial intelligence into vehicle production, particularly in the assembly of electric vehicles. One of the robots currently being tested is called AEON, a humanoid machine developed by Zurich-based Hexagon Robotics.

AEON stands around 1.65 meters tall and weighs about 60 kilograms. Instead of traditional legs, the robot moves using wheels, which allows it to travel quickly and efficiently across the factory floor. BMW is using the robot to assist with tasks such as battery assembly, tightening screws, and handling parts used in high-voltage EV batteries. These are repetitive or physically demanding jobs where robotic assistance can help improve efficiency.

The robot is equipped with 22 sensors, including cameras and microphones, enabling it to analyze its surroundings and avoid obstacles. It also runs on NVIDIA Jetson Orin computing hardware, which processes data and powers its AI-based motion system. BMW describes this approach as “Physical AI,” combining digital intelligence with machines that can interact with the real world.

The project follows a 10-month pilot test in BMW’s Spartanburg factory in the United States, where robots handled more than 90,000 parts and helped produce around 30,000 vehicles.

BMW plans to continue testing the technology, with more trials scheduled soon and a larger pilot program expected to begin later this summer.

Did You Know?

Humanoid robots could soon become common in factories. Companies like Tesla, Mercedes-Benz, and Hyundai are already testing human-shaped robots for assembly-line tasks. Tesla’s Optimus robot, for example, is being developed to handle repetitive factory work and could eventually cost less than a car, making large-scale robotic workers a real possibility in manufacturing plants.

Read More:

• OnePlus Nord 6 Leak Reveals Snapdragon 8s Gen 4, 165Hz Display, and Possible 9000mAh Battery
• ASUS Launches XG27UQDMS: 27-Inch 4K 240Hz QD-OLED Gaming Monitor
• Xiaomi’s New REDMI Smart TV A50 Debuts: 50-inch 4K 144Hz at an Ultra-Low Price

(via)

The post BMW Starts Using Humanoid Robots in Factory to Build Electric Cars appeared first on Gizmochina.

On March 1, 2026, Samsung announced its plans to turn all of its global production facilities into what it calls “AI-driven factories” by 2030. It’s a bold target, and it goes well beyond the traditional automation that already defines modern electronics manufacturing. This time, Samsung is talking about systems that can make decisions on their own.

A key part of that vision is agentic AI, which, unlike standard AI tools, can independently plan and execute tasks in order to meet specific goals. Samsung first introduced the term on the consumer side with the Galaxy S26 series. Now it wants to bring similar decision-making capabilities into its factories to increase efficiency and reduce operational costs over time.

Samsung says it will gradually expand its use of industrial robots, including humanoid models. The company describes three categories: operating robots for managing production lines and facilities, logistics robots for transporting materials, and assembly robots for precise manufacturing tasks.

Humanoid robots are expected to handle more complex or flexible operations, the kind that still benefit from human-like dexterity. Samsung has been building up its robotics capabilities in recent years, including increasing its stake in Rainbow Robotics, which is now part of the group. One of the systems likely to appear on production floors is the RB-Y1, a wheeled humanoid robot equipped with dual arms.

Samsung isn’t alone in exploring this direction. Other manufacturers are experimenting with humanoid robots and AI-powered automation as labor costs rise and supply chains grow more complex. Still, making these systems reliable and economically viable across dozens of global plants is no small task.

For now, Samsung is outlining a roadmap rather than showing a finished product. Whether its factories look radically different by 2030 will depend on how smoothly these technologies integrate into real-world production environments. The ambition is clear. The execution, as always, will be the harder part.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source: Samsung)

The post Samsung plans transition to “AI-driven factories” with humanoid robots and agentic AI by 2030 appeared first on Gizmochina.

Training large language models is brutally expensive. It’s not just about having more GPUs; it’s about how efficiently you use them. And as models scale up, even small inefficiencies can turn into massive time and energy costs.

Now, a team of researchers from MIT, working with collaborators including NVidia, says it has found a surprisingly practical way to reclaim wasted compute during training — in some cases cutting overall training time nearly in half.

The problem they’re targeting lies in reinforcement learning (RL), particularly during what’s known as the “rollout” phase. This is the step where a model generates multiple candidate responses so it can learn which behaviors lead to better outcomes. It’s essential for reasoning-focused LLMs — but it’s also slow.

In fact, the rollout stage can account for as much as 85% of total execution time. The culprit is something researchers call a “long-tail distribution” of response lengths. Most generated responses finish quickly. But a small number run much longer than average. Because GPUs need to synchronize, the faster ones often sit idle waiting for the stragglers to complete.

The MIT team’s solution, called Taming the Long Tail (TLT), tackles that waste head-on. Instead of letting GPUs sit idle during those long generations, TLT uses that downtime to train a lightweight “draft” model on the fly. This smaller model learns continuously from the main model as training progresses.

The idea builds on speculative decoding, a technique where a smaller model predicts tokens ahead of the main model so multiple tokens can be verified in parallel. Traditional speculative decoding relies on a fixed draft model, which quickly becomes outdated as the primary model evolves during reinforcement learning.

TLT changes that dynamic. By retraining the drafter opportunistically using otherwise idle resources, the system keeps the draft model aligned with the main model, without requiring extra dedicated compute.

In experiments across several reasoning-focused LLMs and real-world datasets, the results were significant. The researchers report end-to-end training speedups ranging from 70% to 210% compared to strong baselines, effectively doubling training speed in many scenarios. Importantly, model accuracy remained unchanged.

There’s also an interesting side benefit: the continuously trained drafter itself becomes a useful artifact. Because it’s trained alongside the main model, it can serve as an efficient inference model in certain contexts.

The work points toward a broader theme in AI research right now: optimization over brute force. Instead of scaling up clusters indefinitely, researchers are increasingly looking for ways to extract more performance from the hardware already in place.

If approaches like TLT prove robust at larger industrial scales, they could meaningfully reduce both the financial and environmental costs of training next-generation reasoning models.

Don’t miss a thing! Join our Telegram community for instant updates and grab our free daily newsletter for the best tech stories!

For more daily updates, please visit our News Section.

(Source | Via)

The post Researchers double AI training speed just by reclaiming idle GPU time appeared first on Gizmochina.