AI news, Nov 9, 2026: Alibaba Qwen Launches Team Agent, Google Unveils New TPU, Nvidia Investigates Hinton
Alibaba released Qwen Office 2.0, featuring multi-user agent collaboration to enhance team workflows. Google unveiled two new TPU chips for confidential computing and model data access, challenging Nvidia's dominance. Nvidia is investigating whether CEO Jensen Huang was targeted by a foreign intelligence agency, raising market concerns about geopolitical risks.
Lead written by AI from the day’s stories
Top stories
- Chips
Google Launches Two New TPU Chips to Challenge Nvidia
Google announced two new TPU chips optimized for confidential computing and broader model information access, addressing AI agent needs. The move accelerates product iteration and aims to challenge Nvidia's dominance in the AI chip market.
iThomeIT之家THE DECODER5 outlets
- Products
Disney and Adobe Partner on Generative AI Tools
Disney and Adobe announced a multi-year partnership to develop proprietary "artist-led" generative AI models to support Disney's production pipeline. The collaboration covers theme parks, cruises, and ESPN, aiming to integrate AI into the entertainment industry.
The VergeTHE DECODERITmedia AI+4 outlets
- Products
Alibaba releases Qianwen Office 2.0 for multi-agent collaboration
Alibaba released Qianwen Office 2.0, introducing its first feature for collaboration between multiple humans and AI agents. The update enhances capabilities for automated tasks and human-AI co-creation, shifting agents from individual to team and organizational use with task assignment and progress tracking.
- Products
OpenAI Launches Control Room for Model Workloads
OpenAI released a central system called Control Room that tracks model workloads across its research infrastructure. It allows rapid shutdown of workloads when safety or security concerns arise, enhancing model governance.
- Models
Baidu Discloses RSI System Used for ERNIE 5.5 Flash Development
Baidu revealed its automated research system, ERNIE Evolve, which is used for developing the ERNIE 5.5 Flash model. The system autonomously proposes research directions and validates results, achieving state-of-the-art performance on benchmarks.
雷峰网量子位MarkTechPost3 outlets
Models6
- Models
ByteDance releases Seedream 5.5 image model
ByteDance launched the Seedream 5.5 image creation model, now integrated into Doubao, Doubao Work, and Jimeng. The update significantly boosts generation speed, improves multi-round editing stability, and adds native support for transparent asset generation. The cost per 2K image drops to approximately 0.28 RMB, less than half of the previous Seedream 5.0 Pro.
雷峰网爱范儿THE DECODER3 outlets
- Models
Tencent releases Hunyuan Hy4 model and tests its Agent capabilities
Tencent released its Hunyuan Hy4 model on Nov 4, with plans to open-source its weights soon. GeekPark's evaluation indicates Hy4 has improved in long-horizon agents and multimodal understanding, entering the first tier of open-source models, though it still lags behind top closed-source global models.
- Models
TokenRhythm releases NeoHorse-1.5-Flash
TokenRhythm released the agent-native model NeoHorse-1.5-Flash, featuring 180B total parameters and 6B active parameters. The model is specifically trained to enhance agent tasks and software engineering, currently in limited beta testing.
- Models
Nous Research Releases Arete-27B Open Model Emphasizing Western Values
Nous Research released 'Arete-27B', an open model based on Qwen3.8-27B designed to enhance Western values. The release drew criticism from some quarters, arguing that it introduces reverse bias.
ITmedia AI+GIGAZINE2 outlets
- Models
Shizuku AI launches Shizuku-Duplex for real-time Japanese dialogue
Shizuku AI launched Shizuku-Duplex, a model for real-time Japanese conversation. It processes data every 80ms and supports responses and web search. The release aims to enhance interactive capabilities of AI characters in Japanese contexts.
- Models
Embodied AI Models Launch APIs for Cross-Platform Fine-tuning
Physical Intelligence, Dream Labs, and Yuandao recently opened APIs for embodied AI models, allowing developers to upload data and fine-tune models. These services support cloud-based inference for robots, indicating that embodied models are gaining cross-platform adaptation capabilities.
Products31
- Products
Goldman Sachs removes Claude Code from bankers and traders
Goldman Sachs announced it is removing Claude Code from the computers of its bankers and traders. The move indicates the financial institution is re-evaluating its policies on employee use of generative AI coding tools. It reflects the strict compliance and security management approach of major financial firms toward AI tools.
- Products
Automattic launches Spacefast for AI app building
Automattic has launched Spacefast, a new platform designed to deploy 'vibe coding' applications. The company aims to compete with rivals like Lovable and address the evolving internet landscape. This move highlights the growing influence of AI in software development workflows.
- Products
Sakana AI developing first mobile app
Sakana AI announced it is developing its first mobile application, 'Sakana AI App,' and is currently recruiting testers. The app will integrate the company's existing services to offer a mobile user experience. This move indicates the AI lab is expanding into the consumer mobile market to broaden its product reach.
ITmedia AI+IT之家2 outlets
- Products
Volcano Engine Becomes Hidden Foundation for AI Phone Assistants
InfoQ Chinese details how Volcano Engine serves as the core technology provider for AI assistants in major smartphone brands. As AI phones rise, the company supports underlying models and architectures, indirectly impacting consumer device AI capabilities and ecosystem strategies.
- Products
Google Launches Gemini Chinese Shopping Feature in Taiwan
Google announced the launch of the Gemini shopping feature in Taiwan, supporting the Chinese interface. The feature allows users to search for products and compare prices via conversational AI. This move aims to provide local consumers with a more interactive shopping assistance experience.
- Products
Open-Source Company Deploys AI Agents for Dev Management
An open-source company with 80k stars introduced AI agents to handle coding, task assignment, and bug checking. This demonstrates deep application of AI in software development.
- Products
OpenAI Images 2.6 supports GIF generation
OpenAI launched Images 2.6, a new image generation model featuring GIF animation creation, SVG output, and editing controls. The update is available on ChatGPT, Codex, and the API, with a claimed 50% speed increase. It supports loop animations with 2 to 16 frames.
- Products
GitHub unveils Rubicon architecture for AI agents
GitHub released its new Git storage architecture, Rubicon, designed for massive AI agent collaboration. Internal tests show it supports 5,000 simultaneous AI agents and 7,200 pushes per second, currently available in a limited private preview.
- Products
Perplexity launches Teammates for multi-agent collaboration
Perplexity launched the Teammates feature on Nov. 4, allowing users to create multiple persistent AI agents. These agents can manage long-term projects, track progress, and collaborate by delegating tasks to one another, targeting the proactive AI agent market.
- Products
Anthropic's Claude Marketplace sees 25+ AI apps
Anthropic stated that its Claude Marketplace has launched over 25 third-party agents and services. This indicates the platform ecosystem is expanding, allowing developers to build more AI workflows.
- Products
ElevenLabs Partners with SoundCloud for Direct AI Music Release
AI voice tech company ElevenLabs announced a partnership with music sharing platform SoundCloud. Users can now directly publish AI songs generated by ElevenMusic to SoundCloud, eliminating the need to download and re-upload, which streamlines the process from creation to public release.
- Products
Baidu Dazi Tops Three Enterprise AI Tenders
Baidu Dazi ranked first in three recent enterprise AI tenders, including projects from CITIC Securities. As a general-purpose agent, its monthly active users grew by 1567.5% quarter-over-quarter, highlighting its market competitiveness.
- Products
iOS 27.3 code reveals Siri personal memory system
Researchers found Apple developing a 'Memory Graph' system in iOS 27.3 that uses AI to organize user data. The feature aims to let Siri remember personal details to provide more contextual assistance.
- Products
Meta rolls out AI age estimation to protect teens globally
Meta announced it will roll out AI age estimation globally to include more teens in protected accounts, even if they report adult birthdays. This move aims to address lawsuits regarding minors and strengthen content restrictions.
- Products
Claude Deletes Professor's Class Notes
University of Chicago Law professor William Baude shared an AI incident where Anthropic's Claude accidentally deleted his year's worth of class notes and course outlines while attempting to update lecture page numbers. The event highlights potential risks of AI agents in file operations.
- Products
ChatGPT Mass Ban Disrupts Paid Users Without Warning
On November 8, OpenAI's ChatGPT experienced a sudden, large-scale account ban event without prior warning, affecting numerous users including Pro subscribers paying $200 monthly. Users reported receiving no advance notice, resulting in inaccessibility to their accumulated chat histories and data.
- Products
Samsung One UI 9.5 upgrades Bixby to an AI agent
Samsung is preparing a major upgrade to Bixby in One UI 9.5, transforming it into an AI agent capable of handling tasks like hotel bookings on behalf of users. Currently in internal testing, the feature is expected to launch with the Galaxy S27 series next year, allowing users to create and manage recurring tasks.
- Products
Samsung Tests Galaxy AI Subscription
Samsung is reportedly testing a Galaxy AI Premium subscription at $4.99 monthly. The paid tier offers higher AI usage limits and converts some features to paid, while free features face usage caps.
- Products
DeepSeek plans to launch an official plugin marketplace
DeepSeek announced that its open-source AI agent environment, Harness, will launch an official plugin marketplace with token incentives. The platform has reached 1.54 million daily active users, following similar moves by OpenAI and Anthropic to build agent ecosystems.
- Products
Tianyi and UBTECH partner to deploy humanoid robots
Tianyi Telecom Terminal and UBTECH signed a comprehensive strategic partnership to drive the large-scale application of bionic humanoid robots. They aim to exceed one million home companion robots and 100,000 humanoid robots within three years, establishing a joint innovation lab.
- Products
Apple reportedly develops Siri AI agent for next year
According to TechNews, Apple is reportedly developing an AI agent feature capable of autonomously executing tasks, expected to upgrade Siri further next year. The move aims to enhance the voice assistant's execution capabilities, allowing it to perform complex tasks independently.
- Products
Alipay shares AI assistant 'Xiaozheng' at Zheshang Conference
At the 8th World Zheshang Conference's AI-themed event, Alipay shared its AI assistant 'Xiaozheng', which uses multi-turn dialogue to understand user needs and supports over 10,000 social welfare services like social security and housing funds. This aims to optimize complex administrative processes and enhance user experience through AI.
- Products
South Korea's KAI Develops Conversational AI for Weapon Logistics
South Korea Aerospace Industries (KAI) is developing a conversational AI agent to query maintenance data via natural language, optimizing weapon system efficiency. The initiative aims to enhance logistics support for naval, land, and air weaponry.
- Products
Tencent Workbuddy apologizes for unannounced DeepSeek discount change
Tencent Workbuddy apologized to users for failing to announce changes to the DeepSeek-V4.1 Flash discount. The company will automatically refund 50% of the credit difference for users active between Nov 5 and 8, with credits expected within 7 business days.
- Products
Tencent Yuanqi Platform Ceases Service
Tencent announced that its agent platform Yuanqi will cease public services on November 9, 2026. Users must back up data by January 2027, as functionality will be migrated to Tencent Cloud ADP and other products.
- Products
Aiello Launches AI Customer Service System Integrating Multiple Channels
Aiello launched an AI voice and text customer service system to help companies define workflows between AI and human agents. The system integrates multi-channel conversation logs, allowing staff to focus on complex cases while handling repeated inquiries efficiently.
- Products
Zyphra Launches Beta of MAIA, an AI Agent Tool for Team Collaboration
Japanese AI development company Zyphra released the beta version of its AI agent tool, MAIA. Designed for simultaneous multi-user team usage, the tool aims to streamline collaboration workflows in group settings.
- Products
Claude Introduces Dreaming Feature to Enhance Memory
Anthropic explained Claude's memory feature, introducing a Dreaming mechanism to consolidate and correct erroneous memories. The update improves information accuracy in long conversations, enhancing the AI assistant's context understanding and continuity.
- Products
Taiwan team launches Core Studio AI for video creation
imCore, a Taiwan-based company, launched Core Studio, an AI platform that integrates script writing, storyboarding, and video generation. The tool includes 24 AI agents and a self-trained image model, going live on November 30.
- Products
Claude Code 'Never Show' Feature Leaks Sensitive Info
Security firm Apiiro found that Anthropic's Claude Code stores sensitive information, such as passwords, in configuration files when the 'Never show' option is selected. Over 2,539 secrets were exposed in GitHub repositories, highlighting security risks in AI coding tools.
- Products
Microsoft Launches Guardian Agents to Monitor AI Safety
Microsoft warns enterprises against blindly trusting AI agents, which may cause risks due to misjudgments. The company introduced Guardian Agents, a security mechanism to monitor agent behavior and ensure compliance with organizational rules, balancing AI expansion with safety controls.
Research14
- Research
Tencent AI Chief Warns of AI Bioweapons, Argues Against Closed-Only Approach
Tencent Chief AI Scientist Yao Shunyu warns that AI will soon be able to develop bioweapons, advocating a middle ground between open and closed models. He notes that safety filters are easy to strip while closed models over-block; Tencent's upcoming open-weight model Hy4 will filter dangerous knowledge from training data.
- Research
Sony AI research analyzes IP guardrails in image models
Researchers from Sony AI published a paper evaluating how 14 image-generation models handle existing intellectual property. The study tested the effectiveness of safety guardrails in preventing the generation of protected characters, offering technical insights for AI content compliance.
- Research
Study: ChatGPT holds over 81% of global AI chat traffic
A new study by Graphite shows ChatGPT accounts for over 81% of global major AI chatbot traffic, far ahead of competitors. Data indicates that ChatGPT's monthly visits have resumed rapid growth after a period of stagnation.
- Research
Ant Lingbo Embodied AI Challenge: 15 Teams Advance to 48-Hour Final
The first Ant Lingbo Embodied AI Model Challenge concluded its preliminaries, with 15 teams advancing to the finals. The final round will be held in Shanghai, where teams will compete in a 48-hour physical robot test.
- Research
Weekly AI Time Savings for Developers Drop Sharply
A report by InfoQ Chinese states that the percentage of developers saving several hours weekly using AI dropped from 68% a year ago to 26% today. This shift reflects changing perceptions of AI efficiency and potential challenges in workflow integration.
- Research
IDC: 2026 Global Dexterous Hand Shipments to Reach 25,000
IDC forecasts 25,000 dexterous hand shipments in 2026, with Chinese manufacturers accounting for over 85%. As a key component of embodied AI, its development relies on synergy with AI models.
- Research
DeepMind AI agent finds RSA crypto flaws
Google DeepMind released research showing its AI agent found a double vulnerability in legacy RSA encryption. This highlights AI's growing capability in security testing, potentially altering how cryptographic standards are verified.
- Research
Anthropic reports scientists mainly use Claude for data analysis
Anthropic released a study analyzing 207,000 scientific sessions, finding that researchers primarily use Claude for data analysis and writing. Computer science shows the highest usage, while data collection and hypothesis generation are less common.
- Research
AI-Driven Zero-Day Discovery and Root Privilege Escalation: Linux Security Risks Rise
Research indicates that AI can discover a zero-day vulnerability every 2.5 hours and execute over 400 root privilege escalation attacks, far surpassing human reverse engineering efficiency. This highlights new security threats to Linux systems, as traditional manual audits struggle to keep up with the volume of AI-generated vulnerabilities.
- Research
Y Combinator shows AI agents stuck in bureaucracy in Minecraft civilization
Y Combinator and Olam Labs demonstrated an experiment where 50 AI agent groups build a civilization in Minecraft. The study found the agents suffered from coordination failures resembling 'bureaucracy', stalling the project. This highlights limitations in social interaction and decision-making for large-scale AI agent systems.
- Research
DeepMind applies AI to antibiotic resistance
Google DeepMind is applying AI technology to tackle antimicrobial resistance (AMR), a critical medical challenge. The company is using AI models to analyze resistance mechanisms, aiming to address the global threat of drug-resistant bacteria and improve medical response capabilities.
- Research
Brown University solves multiferroic chip miniaturization issue
A Brown University research team announced a solution to a long-standing issue preventing multiferroic materials from entering ultra-low-power chips. This finding aims to enable the miniaturization of such materials, potentially driving the development of low-energy computing.
- Research
PSI Confirms AI Solves Quantum Computing Math Problem
AI startup Physical Superintelligence announced it proved the Aaronson-Anthony conjecture using an Anthropic model. The proof was verified by Lean, resolving a 2008 unsolved quantum computing problem.
- Research
Linux Foundation: AI boosts speed but floods open source with 'slop'
A study by the Linux Foundation and Red Hat found that while generative AI halves coding time, it increases review pressure for maintainers. Roughly 25-31% of submissions are identified as low-quality 'AI Slop,' prompting calls for new guidelines to manage code quality.
Open source5
- Open source
LG AI Research Opens K-EXAONE 2.5 API for Free
LG AI Research has made the K-EXAONE 2.5 model API available for free to developers and enterprises. The model, which underwent additional training, exhibits agent capabilities comparable to leading open-source models like DeepSeek, scoring competitively on various benchmarks and facilitating enterprise AI adoption.
IT之家NVIDIA Blog2 outlets
- Open source
Reason Machines Open-Sources Aspen, a Git Server for AI Agents
Reason Machines has open-sourced Aspen, a self-hosted Git server designed for AI agents. It supports the OpenAI Pando design and achieved 567 pushes per second on a single 8-core server, enhancing AI development infrastructure.
- Open source
0sec Labs Open-Sources AI Security Harness 0
0sec Labs has open-sourced 0, an AI security harness for code review, pentesting, and reverse engineering. It supports execution via browser, CLI, and MCP server, allowing users to customize model providers to enhance local security defenses.
- Open source
Meituan open-sources LongCat-2.5 model
Meituan LongCat open-sourced its native multimodal model LongCat-2.5 under the MIT license. The model has 1.6 trillion parameters, supports 1M context, and is optimized for domestic compute chips to enhance agentic and visual coding tasks.
- Open source
Linum releases low-cost image generation model P-JiT
Linum, an AI research organization by two researchers, released the image generation model P-JiT. The model aims to address high generation costs and poor controllability, developing for approximately $21,000. It is now available as open weights.
Funding12
- Funding
NVIDIA Announces Investment in Korean AI Startup Point2
NVIDIA hosted an AI startup forum with Korea's MSIT and announced an investment in Korean startup Point2. The collaboration spans foundation models and AI infrastructure, aiming to advance South Korea's physical AI ecosystem.
- Funding
Microsoft folds AI Frontiers team, head exits
Microsoft Research has closed its AI Frontiers team, merging members into the Microsoft AI division. Head John Langford announced his departure, describing the project's failure.
- Funding
Dropbox acquires Tigris Data for $350M to expand into AI storage
Dropbox has announced a $350 million acquisition of AI object storage startup Tigris Data and the launch of DBX Agent Cloud. This move marks Dropbox's expansion from traditional file storage into the AI infrastructure market, providing data storage services for enterprise AI applications and agents.
- Funding
AI Startup Coframe Completes Series A Funding
AI startup Coframe has raised $25 million in Series A funding, bringing its total to $35 million. The company aims to use AI to reshape the internet by providing personalized web pages for every visitor.
- Funding
Sakana AI Plans 2027 IPO
Sakana AI plans an IPO by fall 2027 with Goldman Sachs as advisor. The company is also developing the Sawara model series, aiming to be Japan's first listed GenAI lab.
- Funding
Pudong VC links AI startup training with seed funding
Pudong VC concluded its AI training camp in Zhangjiang, signing all 20 early-stage teams. Using a RMB 2 billion AI seed fund, the program offers immediate seed funding upon entry and up to RMB 1.5 million in follow-on investment to support nascent AI startups.
- Funding
Anthropic Orders $3.4B in Gas Turbines for AI
Anthropic reportedly spent $3.4 billion on a 1.7GW gas turbine order to secure power for AI compute. This is the largest known power equipment purchase by an AI company, signaling infrastructure investment into power generation.
- Funding
Musk projects SpaceX annual revenue to exceed $500B by end of next year
SpaceX CEO Elon Musk projects that driven by AI computing, the company's annualized revenue will exceed $500 billion by the end of 2027. This expectation highlights AI computing as a key driver of rapid revenue growth for SpaceX.
- Funding
Analog Devices acquires Alif Semi for AI edge applications
Analog Devices completed the acquisition of Alif Semiconductor, integrating its AI-native platform. The move aims to expand the smart edge market and accelerate application innovation in industrial, robotics, and digital health sectors.
- Funding
Microsoft reportedly cancels Oracle AI compute lease, causing stock volatility
Reports indicate that Microsoft has not renewed its AI computing resource lease with Oracle and is reassessing a multi-billion dollar cloud contract. The news triggered significant volatility in Oracle's stock, reflecting market concerns about the sustainability of AI compute demand.
- Funding
CoreWeave seeks cooperation with Texas
AI cloud provider CoreWeave has written to the incoming Governor of Texas to seek cooperation in AI infrastructure development. Changes in the state's political landscape could impact its AI infrastructure prospects, and CoreWeave aims to build a partnership with the new administration to advance its operations.
- Funding
42AI raises $62.5M seed round for behavioral AI in finance
Israeli startup 42AI has ended its stealth mode and announced the completion of a $62.5 million seed funding round. The company focuses on leveraging behavioral AI technology to enter the financial services sector.
Policy14
- Policy
Trump Says AI Risks Are Manageable, DOJ Involvement Claimed
In a 60 Minutes interview, Trump responded to AI executives' safety warnings, stating that risks are not as severe as portrayed and are manageable. He claimed the Department of Justice is overseeing these risks, asserting that AI safety is under control.
- Policy
Elvis and Eminem rights holders sue Suno for infringement
Rights holders for Elvis Presley's music and the publisher of Eminem's records have sued AI music company Suno. They claim Suno illegally generates music and lyrics mimicking these artists. The lawsuits highlight legal risks regarding copyright infringement in the AI music industry.
- Policy
Indonesian residents protest AI data center water depletion
Residents of Batam Island, Indonesia, are protesting as the construction of multiple data centers supporting the global AI boom has caused local tap water supplies to deplete. Daily activities like bathing and cooking have become difficult due to severe water shortages, sparking public outcry over resource allocation.
- Policy
South Korea Bans AI Glasses in College Entrance Exam
South Korea's Ministry of Education has mandated a ban on AI glasses and other electronic devices in the upcoming college entrance exam. Dedicated staff will inspect candidates at each site, and violators will have results voided, addressing recent cheating incidents.
- Policy
Apple Intelligence completes AI filing in China
According to CAC Shanghai, Apple Intelligence has completed filing for tablet and laptop scenarios, indicating the feature may soon launch on mainland China iPad and MacBook models. This filing aligns with previous iPhone model registrations, marking a key compliance step for Apple's AI services in the region.
- Policy
Norway Halts New Data Center Grid Access
Norway will restrict the expansion of energy-intensive data centers, with grid operator Statnett suspending capacity reservations for new projects. A new licensing system will prioritize critical infrastructure like defense and healthcare to address potential power shortages and rising electricity prices.
- Policy
SoftBank considers shifting AI data center growth to other countries
SoftBank Corp. CEO Junichi Miyanishi stated that the company is reassessing its US market prospects due to the Democratic midterm election victory and rising local opposition to data centers. If the US situation worsens, SoftBank may expand its AI cloud computing business in other countries, with current cloud and AI revenue up 37%.
- Policy
Gartner: Agent Governance as 2027 Trend
Gartner released its 2027 top 10 trends report. It highlights that as AI becomes smarter, enterprise management becomes harder, making agent governance a new focus. This reflects a shift in AI governance policy.
- Policy
Google's UK data center plan faces opposition from 55,000 petitioners
Nearly 55,000 people signed a petition asking Google to abandon its plan to build a large data center in Thurrock, UK. Opposing groups, including wildlife organizations and a motorcycle club, raised concerns about habitat destruction, pollution, and the loss of event spaces, sparking environmental and community conflicts.
- Policy
China launches unified AI and algorithm filing system
China's Cyberspace Administration launched the Internet New Technology and Application Information Service Filing System on Nov 9. The system integrates filing for generative AI, algorithms, and deep synthesis, replacing previous systems to improve standardized management.
- Policy
DARPA Launches AI for Market Resilience Program
The US Defense Advanced Research Projects Agency (DARPA) launched the "Intrinsic Market Resilience" initiative to use AI for identifying design vulnerabilities in electronic markets. The move aims to strengthen financial market defenses against hostile actors attempting to disrupt the economy.
- Policy
Trump says AI risks manageable, open to regulatory guardrails
US President Trump stated that AI risks are manageable but expressed openness to working with Democrats on regulatory measures. His shift from supporting industry self-regulation to accepting government guardrails marks a potential change in policy stance, though the extent remains unclear.
- Policy
NYT: Companies Approved to Handle Nvidia H200s Linked to Shell Firms
The New York Times investigation revealed that some Chinese companies approved by the US Department of Commerce to handle Nvidia H200 chips have suspicious registered addresses, potentially being shell companies. This raises concerns about AI chip export controls.
- Policy
AI Hackers Attack Japanese and Korean Firms
AI evolution changes cybercrime, with automated attacks increasing security risks. Japanese and Korean firms face threats from automated intrusions, impacting operational security.
Chips3
- Chips
Microsoft reportedly ends Oracle AI compute lease
Microsoft reportedly decided not to renew its 2023 compute lease with Oracle for AI-powered Bing. The capacity, using older NVIDIA chips, has been shut down recently, reflecting shifts in cloud compute demand.
- Chips
DEEPX Launches Low-Heat AI Chip DX-M1
South Korean AI chip startup DEEPX unveiled the low-heat AI chip DX-M1 in the US. The CEO emphasized that its thermal performance is significantly better than standard products.
- Chips
AI Demand Causes Memory Shortage to Persist Until 2030
Rising AI data center demand has tightened global memory supply, with AI consuming about 70% of DRAM. Analysts predict the shortage will last into the 2030s, keeping GPU and hardware prices high.
Other11
- Other
Nvidia investigates targeting of Jensen Huang by foreign intelligence agency
Nvidia is investigating whether CEO Jensen Huang was targeted by a foreign intelligence agency. The security concerns sparked market worries, leading to a post-market drop in Nvidia's stock. The incident highlights geopolitical and intelligence risks faced by tech giants, potentially affecting investor confidence in its long-term stability.
科技新報THE DECODERArs Technica3 outlets
- Other
NVIDIA helps advance South Korea's physical AI ecosystem
NVIDIA is helping South Korea apply AI to semiconductor manufacturing, robotics, and autonomous vehicles. These efforts use accelerated computing and simulation to move physical AI from development into real-world production.
NVIDIA BlogIT之家2 outlets
- Other
Apple caps engineers' daily AI budget at $200
Apple has reportedly implemented a daily cap of $200 for engineers' AI tool usage. This move reflects how major tech companies are balancing innovation with cost control when adopting generative AI.
- Other
Hyundai and Korean firms adopt NVIDIA platforms for secure AI agents
Hyundai Motor Group and AIM Intelligence are integrating NVIDIA's secure AI runtime environments. These companies leverage open models and security architectures to enhance the stability and safety of AI agents in enterprise workflows.
- Other
NVIDIA and South Korea expand sovereign AI collaboration
NVIDIA and South Korea's MSIT have expanded their collaboration to cover AI chips, data center infrastructure, and talent development. The partnership aims to build a sovereign AI ecosystem and explores opportunities for NPU and XPU chip companies to join the NVLink Fusion ecosystem.
- Other
Stanford CS professor shares AI-native engineering pedagogy
Mihail Eric shared Stanford's CS146S course experience at Monaco, explaining how students use AI tools to one-shot previous homework. The talk explored AI-native engineering pedagogy.
- Other
Kaohsiung Police Crack Case of Man Using AI to Steal Data
Kaohsiung police solved a case where a 38-year-old man used AI to learn coding and build automated attack systems, stealing over 230,000 personal records for sale. The suspect was arrested and referred to prosecutors on charges of computer misuse.
- Other
Acronis Reveals BossAI Targets TW Emergency Services
Security firm Acronis revealed BossAI, an automated AI attack system created by a Taiwanese actor. Comprising 12 AI agents, it targeted hundreds of organizations in 11 countries over 12 days, with Taiwan's emergency services hit most heavily.
- Other
Alibaba Cloud Wins China Patent Gold Award for AI Compute Scheduling
Alibaba Cloud's patent for dynamic AI compute scheduling won China's highest patent award. The technology addresses elasticity and security challenges in large-scale AI infrastructure, highlighting the company's innovation in AI compute management.
- Other
Pierre Fabre Partners with XtalPi on AI Drug Discovery
Pierre Fabre and XtalPi signed a collaboration to discover anticancer drugs using AI and robotics. XtalPi will design small molecules while Pierre Fabre handles clinical development and validation.
- Other
Microsoft files patent for real-time AI live stream editing
Microsoft filed a patent for an AI system that listens to live streams in real time to replace words based on rules. Generative AI then redraws the speaker's lip movements to hide the audio modifications, applicable to live video or AR presentations.
All news → · All rights belong to the original creators