404 Media7575Claude Code v2.1.159 released, GitHub Stars exceed 342k, users utilize AI reverse engineering TV operating systems, smart devices, removing redundant features and ads; this tool can analyze application functions without source code and replicate them, triggering a trend of software cloning and modification, but the quality of the results is uneven.
NVIDIA NewsroomNVIDIA has signed an agreement with IBIDEN, Samsung Electro-Mechanics, and Unimicron Technology Corporation.
The DecoderSymantec researchers discovered an aggressive AI agent that hijacks unprotected GPT/LLM models on the internet, executing attacks without human intervention within 100,000 steps and making the model owners bear the computational costs. The agent scans server vulnerabilities, steals database credentials, installs crypto-mining programs, and uses a five-layer jailbreak strategy to bypass security restrictions. Attackers embed the AI core into attack tools, making language models responsible for planning and dedicated modules executing the attacks, forming a self-sustaining attack loop.
The Verge · AI7070Eiliko's AI robots were exposed to have serious security vulnerabilities, allowing hackers to access all conversation records. Eiliko responded by hiring an independent security company to conduct a comprehensive review of the system and closed the interface for remote modification of robot responses. However, the long-term memory and custom avatar features may still store user data, and the current system still exists security risks, hackers can still access unprotected audio and photo records.
17-year-old AI researcher Chen Guangyu leaves Moonshot AI to join Jeff Dean's Discovery Loop team, having worked there for a year and thanking Du Yulun for being hired at the age of 16
The final round of the Lark Beanpack Work AI Pioneer Future Talent Competition was held in Beijing, where 12 teams used Lark and Beanpack Work to solve real business problems for companies, with entrepreneurs, business leaders, and university scholars present to evaluate the results.
OpenAI acquires AI biopharmaceutical company Orchestra Bio and integrates its team and technology into its life science AI model Rosalind. Orchestra Bio's CTO is a Peking University alumnus.
OpenAI GPT-5.6 Sol identified a critical vulnerability in the open-source programming language PHP. Over the past two months, OpenAI led a coalition of maintainers, researchers, and security providers to address the vulnerability. PHP powers approximately 70% of websites worldwide.
How the accounts differ:TechCrunch states the vulnerabilities affect '70% of websites'; The Decoder and Latent Space state 'affecting millions of websites'.
PyTorch BlogNscale announced its joining the PyTorch Foundation as a platinum member, with its full-stack AI cloud platform integrating data centers, computing, and software. Nscale plans to continue investing in Ray's development and contributor community after acquiring Ray's parent company, Anyscale. Mark Collier, executive director of the PyTorch Foundation, stated that Nscale's joining will optimize multi-node performance and ensure seamless operation of PyTorch on advanced hardware.
The final round of the Lark Doubao Work AI Pioneer Future Talent Competition was held in Beijing on October 20, with 12 teams using Lark and Doubao Work to solve real business problems for enterprises. The competition opened all Lark AI functions to contestants and provided unlimited Token benefits, with over 80% of Token usage occurring in Doubao Work. Enterprises select talent through project works rather than relying solely on resumes.
Sony has released 4 updates in the past 5 weeks to patch PS5 exploits, firmware 14.20 fixes a security flaw targeting the ARM Cortex-A53 coprocessor, which allows attackers to execute code in the coprocessor firmware to bypass security checks. Previous firmware updates released in September, 14.00 and 14.10, have created a version wall and pushed the jailbreak ecosystem forward.
PanLin Technology released the Q1 financial report for the 2027 fiscal year, with net profit attributable to shareholders of $2.753 billion, a year-on-year increase of 75.51%, and revenue of $8.219 billion, a year-on-year increase of 54.38%, slightly better than market expectations. Net cash flow from operating activities was $3.498 billion, a year-on-year increase of 96.61%.
Zhiyuan Robotics broke the record for office building leasing in Shanghai, signing the Qiantan Yuan, Qiantan Shanghe, and Yiqiao Smart Valley projects under Lujiazui Group, with a total area exceeding 220,000 square meters, and an annual rent of about 350 million yuan. Zhiyuan was founded in 2023, with embodied intelligence as its core technology, breaking the global record for mass production of embodied intelligence robots. In the first half of 2026, it shipped 9,700 units, ranking first globally with a market share exceeding 43%.
Yushu Technology clarified its future three major layout directions: enhancing the mass production capability of robots, advancing the development of physical AI small models through self-research and cooperation, and laying out ultra-large-scale computing power clusters. At the same time, it officially announced the World Physical AI Competition, investing tens of millions of yuan in annual prizes, opening eight tracks to global teams and individuals, aiming to push the boundaries of physical AI technology.
Samsung Electro-Mechanics signs a five-year FC-BGA substrate supply agreement with Tesla for Tesla's electric vehicles and Optimus robot AI chips, Tesla provides approximately 1.1 trillion KRW in production expansion funding, Samsung's Seongnam factory facilities are expected to be delivered in 2029.
MIT Technology Review · AI7575In corporate AI inference costs, data storage overhead is often overlooked. After each model response, a large amount of backend data needs to be stored, including logs, caches, backups, etc., which is 130 times the size of the output files users see. McKinsey predicts that by 2030, AI inference demand will account for 40% of total data center demand, and nearly 95% of IT decision-makers report that data volume has increased by generative AI in the past year.
Latent SpaceSame story · 4 outlets00OpenAI's Patch the Planet project with security company Calif and Wiz jointly discovers PHP2Shell vulnerability affecting 70% of websites, respectively using GPT-5.6 Sol and Gemini 4 Argon for discovery and repair; the two parties coordinated and repaired it about two months before the vulnerability disclosure, Cloudflare deployed WAF mitigation measures, NSA recommends immediate system updates.
How the accounts differ:TechCrunch states the vulnerabilities affect '70% of websites'; The Decoder and Latent Space state 'affecting millions of websites'.
NVIDIA BlogNVIDIA collaborates with governments in multiple European countries to apply advanced models to public management while ensuring control over data and infrastructure. The collaboration projects include: Germany developing an intelligent platform based on a localized dataset Nemotron-Personas-Germany.
OpenAI News4040Cloudflare uses OpenAI Daybreak to investigate vulnerabilities and propose fixes for review, with developers using Codex to build faster, higher-quality code.
TechCrunch · AISame story · 4 outletsHow the accounts differ:TechCrunch states the vulnerabilities affect '70% of websites'; The Decoder and Latent Space state 'affecting millions of websites'.
TechCrunch · AI6565Blackstone executive Jas Khaira says concerns about AI infrastructure debt are exaggerated. Blackstone is participating in the AI boom through massive lending, data center investments, and investments in cutting-edge AI models and applications. Its AI program center BXN1 has supported computing-related transactions like Crux AI and pioneered a financing model based on customer contracts and GPUs as collateral. Khaira notes that while rising interest rates and inflation are increasing borrowing costs, the risk of AI infrastructure debt is manageable because most data centers serve creditworthy super-large enterprises.
The Verge · AI4040Meta announced the issuance of a class of non-voting stock to fund its AI initiatives, with the transaction set to begin in 2027. The company expects its capital expenditures this year to be between 135 billion and 145 billion US dollars. Muse AI agents were launched this week, but their conversation sharing feature has been paused due to privacy issues.
The DecoderFeaturedSame story · 4 outlets7878OpenAI's GPT-5.6 Sol and Google's Gemini 4 Argon discovered a critical PHP vulnerability affecting millions of websites in the Patch-the-Planet program, which was independently discovered by the security company Wiz using its own Wiz Atlas platform.
How the accounts differ:TechCrunch states the vulnerabilities affect '70% of websites'; The Decoder and Latent Space state 'affecting millions of websites'.
Why it matters: The incident involved collaboration among multiple institutions to fix the issue, and also demonstrated the dual role of the model in both vulnerability discovery and attack amplification.
OpenAI News5050AT&T collaborates with OpenAI to reshape legal knowledge acquisition and case intake processes through the LegalEdge team, enhancing legal judgment capabilities.
Ars Technica · AISame story · 2 outlets2828SpaceXAI is being sued by the NAACP for unauthorized gas turbines, facing fines of up to $124,426 per day. Trump attempted to intervene, claiming the EPA has the authority to unilaterally dismiss the lawsuit, but the judge rejected this claim.
How the accounts differ:Ars Technica states SpaceXAI faces a fine of up to 124,426 USD/day, while TechCrunch does not mention the fine amount.
SemiAnalysisDeepSeek released the Ascend support for TileLang, DeepGEMM, DeepEP, TileKernels, FlashMLA, and DeepSelect before the National Day holiday. Huawei provides hardware and native software stack, while DeepSeek contributes workload and engineering capabilities, complementing each other to optimize model performance.
Anthropic Newsroom5555Experimental physicist Christopher Stubbs joins Anthropic's long-term benefit trust, he will use decades of cosmology observation and precision measurement science expertise. Stubbs is a professor at Harvard University and has served as the Dean of the Harvard School of Science. In 1998, he participated in the discovery of the universe's accelerating expansion and won the Breakthrough Prize. Anthropic co-founder and President Daniela Amodei said that Stubbs's rigor and pursuit of truth will shape the future of AI technology.
Ars Technica · AI6565OpenEvidence focuses on literature-based drug discovery, with its economic value expected to far exceed charges to hospitals or advertisers. Tarashansky, the creator of Anthropic platform Claude Science, has joined the company to support clinical work.
Newcomer3535MVP Ventures is negotiating to raise 100 million US dollars to increase its stake in companies such as Anduril, Erebor, and ModRetro, all founded by Palmer Luckey.
Recode China AISame story · 2 outletsThe draft export control regulations issued by China's Ministry of Commerce and Ministry of Science and Technology for the first time cover advanced large AI models, AI agents, and large-scale computing clusters, with a total of 40 banned and restricted technologies, including models trained using China's legal 'core data'.
How the accounts differ:Recode China AI states the ban covers 40 items, while Hacker News specifies 23 items as AI-related.
404 Media1010A large data center threatens the endangered chestnut forests of the United States, which were the subject of many works by the artist Alfred Stieglitz.
ChinaTalkChina's Ministry of Commerce and Ministry of Science and Technology propose adding more AI technologies to the export control list, including cutting-edge models, AI chip design, memory chips, and clusters. Some are banned for export, while others require licenses, and all new items must undergo pre-export registration before the list takes effect. This move aims to protect intellectual property and national security but may impact the global openness and competitiveness of China's AI industry.
OpenAI NewsFeatured6262OpenAI Collaborates with Google Cloud, Integrating OpenAI Frontline Models and Codex Workflows into Gemini's Model Garden and Offering Through Google Cloud Marketplace
Why it matters: OpenAI's cutting-edge models and Codex workflow enter Google Cloud's Model Garden, providing a new entry point for cross-cloud distribution for readers to explore.
