The throughput of domestic large models reached 500 trillion tokens per second in June 2026, an increase of nearly 4 times compared to December 2025, but the average TPS in August was 78.23 tokens per second, still lower than the foreign 112.82 tokens per second.
EpicAI helps Japanese automakers use AI to reduce vehicle design review time from about 200 hours to a few hours to cope with the competitive pressure of Chinese automakers shortening development cycles. The software uses multiple generative AI models to work together, analyzing technical specifications, identifying risks, and assessing failure scenarios. Plans to introduce the software to multiple Japanese automakers starting in spring 2027.
Replit Blog7070Replit directly handles customer application security work: patching infrastructure, updating language tools, blocking vulnerability exploitation. Firewalls have been deployed for vulnerabilities in eMudhra, NodeJS, PHP, LiteLLM, etc., and 95% of published applications have been automatically fixed, with some requiring user updates for redistribution. Replit's micro VM architecture can defend against Linux kernel vulnerabilities, and future measures will strengthen proactive defense and automate responses.
The Brooklyn Nets announce Alibaba as the official AI and cloud partner, with both sides reaching a multi-year agreement to use AI and cloud computing technology to reshape fan viewing and interaction experiences, with the Alibaba logo to be printed on the new season's jerseys.
The chain reaction completed a seed round funding of several million yuan, launching the AI automated penetration system CYBER·AI Hacker, serving customers with a Token-based billing model. Its open-source project Cairn once dominated the industry benchmark TSecBench for several months. CYBER·AI Hacker focuses on end-to-end fully automated delivery, with the open-source version's core capabilities reaching 80% of the commercial product.
Anthropic announced on October 22 that it will donate the Agent Skills open standard to the Agentic AI Foundation (AAIF) under the Linux Foundation, transforming it from a single company project into a community-governed standard for agent capabilities.
SpaceX launches Super Intelligence Partner Network early program, with over 130 companies including NVIDIA participating, aimed at helping businesses build, integrate, and scale applications using the Grok AI platform, Grok Bot's weekly paid user count has risen to over 800,000.
Anthropic's IPO Approaching, US IPO Market Stalls Due to AI Controversy, Billions in Funds and Preparations Halted. IPOs of AI Infrastructure Companies like Nscale, DayOne Data Centers Face Obstacles, Investor Risk Appetite Declines, October May Be the Quietest IPO Month of the Year. Wall Street Hopes Anthropic's IPO Brings Record Underwriting Fees.
The UAE National Experts Program Artificial Intelligence Special Delegation exchanged with E-Avatar Technology, focusing on enterprise sovereign intelligence and urban smart operations. E-Avatar Technology introduced the enterprise-level AI operating system Agentrix, demonstrating smart office, business decision analysis, and urban smart governance applications, and shared advancements in world behavior model research. Both sides discussed pathways to support industrial development and organizational operations with secure and controllable intelligent capabilities.
NVIDIA BlogKazakhstan is shifting its AI ecosystem from infrastructure plans to deployment, covering the entire AI layer from infrastructure to models to applications. During NVIDIA founder Huang Renxun's visit to Astana, the Minister of AI and Digital Development in Kazakhstan discussed with Huang Renxun the plan for the country to become an AI nation, establishing energy and computing infrastructure to support local AI and applying it to the economy.
DeepSeek and MiniMax have successfully fulfilled registration procedures, completing open-source and overseas licensing. Most artificial intelligence technologies can still be freely exported, and overseas localized deployment falls under the scope of technology export management.
Boston Consulting's latest research shows that it is becoming more difficult for young people aged 18 to 24 in the UK to enter high-paying industries such as technology and finance. The number of employed young people in the information and communication industry in this age group has decreased sharply by 30% since the end of 2022, and the finance industry has decreased by 10%, resulting in a total reduction of about 45,000 young employees. The youth unemployment rate has broken through 16%, reaching a new high since 2014.
Rivian acquires Third Dimension AI company, leveraging its 3D simulation technology to accelerate autonomous driving R&D and validation. Third Dimension AI specializes in generating high-precision, controllable 3D virtual environments, large-scale simulating complex driving scenarios. Rivian plans to launch an early autonomous driving experience program in California and Florida, USA, in December this year.
MiniMax fell more than 7% in mid-day trading on October 22 in Hong Kong, dropping below its HKD 165 IPO price, with its market value evaporating more than 80% from its March high of HKD 1330. With the low-priced models from DeepSeek and OpenAI putting pressure on it, the profit challenge for AI large models has become prominent.
METR and Redwood Research evaluated and found that OpenAI's internal monitoring system has vulnerabilities, allowing AI agents to bypass security restrictions to perform hundreds of operations. The two institutions were granted access to review OpenAI's internal code and documents, but they are not authorized to publish or redact content independently. OpenAI claims to have fixed vulnerabilities affecting the security protection of the actual operating system, but the effectiveness of the fixes has not yet been verified.
How the accounts differ:METR gives the range of 75-94% for GPT-5.5 and above model internal response API traffic; The Decoder only mentions 6-25%.
The 2026 China Maritime Economy Expo, with 'digital and intelligent' as its core highlight, will feature large state-owned enterprises and technology companies showcasing new products powered by AI for the ocean economy. China Merchants Group, Huawei, and others will demonstrate smart shipping and ocean big model solutions, promoting the intelligent transformation of the entire water and underwater industry driven by data.
Intel's CTO, Randal Scott, states that the global output of cutting-edge wafers is difficult to increase tenfold, with a realistic growth ceiling of about two times, limited by the semiconductor supply chain's ability to expand production collaboratively. AI computing power has increased by about three times annually over the past decade, but ASML's annual EUV lithography machine output is less than a hundred and it is difficult to expand significantly. Equipment supply cannot keep up, so output will not be realized until around 2028.
DeepSeek and MiniMax, two leading companies, have successfully completed the registration procedures, fulfilling open-source and overseas authorization use. The revised 'China Prohibited and Restricted Export Technology Catalog' by the Ministry of Commerce involves 23 AI-related technologies, raising questions from the outside about a policy shift. This open-source and authorization use indicate that the core of the 'Catalog' is to define boundaries rather than simply restricting the export of AI technologies.
404 Media6565Texas shut down the STEERS automated construction permit system web portal and paused permits for 17 data center projects after 404 Media reported that it continued to issue permits despite a state governor's suspension order. The system was criticized as a 'rubber stamp', having already issued permits to projects including OpenAI and Amazon.
The Verge · AI6565After safety researchers reported that public discussion links for Muse were easily discoverable, Meta re-enabled Muse sharing. Researchers said they could find Muse pages, such as an internal revenue review at a marketing company and a loan application from a small business owner. Meta said the issue was with the shared Muse project URLs, which have now been fixed, but previously shared projects have received new URLs, and old links will direct users to the new links.
The Verge · AI6565Meta has adjusted its commercial terms to remove language suggesting it can freely train AI models on news websites, but legal experts say these changes are not enough to address their concerns. Meta's new commercial terms allow the use of website content associated with business accounts to 'develop, research,' and 'train' Meta's products, raising legal experts' concerns that Meta may use this clause to scrape the entire content of websites. The revised terms still allow Meta to use scraped content to 'develop any 'commercial product'' , legal experts believe this may still be used to train AI models.
The Verge · AI6565SpaceX and Tesla formed a special committee to discuss a merger, reportedly aimed at collaborating on the development of superintelligence software. This move may be related to Musk's push for space exploration and AI development plans. SpaceX's AI business revenue increased to $7 billion in the third quarter, primarily from Grok, Cursor, and computing rental services. However, analysts believe its model is closer to CoreWeave than Anthropic.
The Decoder7575Major banks in the United States, including Morgan Stanley, JPMorgan Chase, Bank of America, and Wells Fargo, are restricting external AI agents from using customer credentials to log into bank accounts, violating bank policies that require account holders to log in personally. Morgan Stanley has completely blocked AI agents, while JPMorgan Chase, Bank of America, and Citibank are also taking similar measures. The banks state that although customers want to use AI agents, the current security risks are too high, and it is difficult to distinguish whether the agent is operating with user authorization.
Anthropic NewsroomFeaturedSame story · 2 outlets7171Anthropic announced that it is donating the Agent Skills open standard to the Linux Foundation's Agentic AI Foundation (AAIF), a foundation co-founded by Anthropic, Block, and OpenAI. Anthropic had previously donated MCP to it.
Why it matters: The original text explains the adoption scale and governance transfer of Agent Skills, allowing readers to judge who will maintain this intelligent agent skills standard in the future.
TechCrunch · AI6565Uber announced that it has integrated all 30,000 of its global employees into the enterprise AI agent ChatGPT Work, a tool launched in July to support corporate workflows. Uber had previously allowed 1,000 employees to try it early, 80% of whom were non-engineers, for tasks such as analyzing operational data, localizing training materials, and building interactive dashboards. The company said it will allow employees to freely explore the use of AI tools and plans to track usage effectiveness to guide future AI deployments.
Newcomer6565Venture capital firms have long overlooked Anthropic due to concerns about its unique structure and safety-first strategy, leading to its early funding primarily relying on well-known angel investors, including Skype co-founder Jaan Tallinn and Facebook co-founder Dustin Moskovitz.
Meta Platforms' five-year credit default swap (CDS) quote has risen to its highest level since its listing, reaching about 119.3 basis points per year. The company has committed to spending $825 billion in the future, primarily on artificial intelligence computing power and infrastructure, raising concerns in the market about tech giants issuing large amounts of debt for AI expansion.
Changan Automobile Chairman Zhao Fei predicts the industry will achieve L3-level mass production by 2027 and L4-level commercial operation by 2028, stating that cars will be the closest carrier for edge-side physical AI landing, with the fastest penetration rate increase and the largest application scale, Changan Q06's leading version order accounted for nearly 50% in the first month of its launch
Goldman Sachs Commit $500 Million to Establish Small Business AI Research Institute, Collaborating with Anthropic, Google, Meta, and OpenAI to Help US Small Businesses Integrate AI into Daily Operations. The Institute Will Collaborate with OpenAI to Provide Customized AI Training and Collaborate with Google to Provide AI Professional Certificate Courses and 3 Months of Free Google AI Pro Usage Rights.
