OpenAI Deployment SafetyGPT-Image-2.6 Sunburst and GPT-Image-2.6 Flare are trained on public data, third-party collaboration data, and user-generated data, employing a strict data filtering process including advanced privacy information filtering and security classifiers to reduce harmful content.
ChatGPT now supports direct generation and editing of GIF animations, this update was launched in sync with OpenAI's Images 2.6 model, allowing users to adjust the speed of GIF animations and edit individual frames. Over 4 billion images are generated weekly through ChatGPT and its API.
Apple has tightened the budget for employees using AI models, with daily spending limits for some engineers using AI models reduced from 300 dollars to 200 dollars, and plans to further cut some engineering teams to 100 dollars, with limits of 100 dollars for some non-engineering teams. Apple employees can use Claude, ChatGPT, and Gemini services.
OpenAI has released the image generation model Images 2.6, which is now open to global users. Image generation speed is 50% faster than the previous version, with a new GIF generation feature added. This model supports use in ChatGPT, Codex, and API. The Flare version can generate images in 6 seconds under low-quality settings.
Google provides a financial guarantee for the partial leasing or performance obligations of Anthropic in TeraWulf's related projects, aiming to reduce project financing costs and accelerate the construction of AI data centers.
Goldman Sachs and Apollo Global Management are in discussions with investors to sell a portion of the $40 billion SpaceX AI chip transaction, having contacted potential buyers and asking some investors to purchase up to $5 billion in shares before the syndicated loan process begins. The transaction aims to avoid debt being recorded on SpaceX's balance sheet, making it the largest GPU-like transaction to date.
Apple has reduced the daily AI spending limit for some engineers, lowering it from 200 dollars to 100 dollars, and is considering cutting the budget for non-engineering teams to 25 dollars; this move aims to control AI usage costs, requiring employees to use services like Claude, ChatGPT, and Gemini, or switch to local AI solutions.
XPeng Group Chairman He Xiaopeng attended the first batch of MONA L05 owner delivery ceremony on November 4th, completing the delivery of new vehicles. MONA L05 offers 6 models with a price range of 14.48-17.68 yuan, and a limited-time promotional price range of 13.98-17.18 yuan. The new vehicle is the first large five-seat SUV in the MONA series, equipped with a pure visual intelligent driving solution. The Max version has an effective computing power of 750TOPS, and the Ultra SE version has an effective computing power of 1500TOPS.
OpenAI's automated vulnerability research chain identified 2308 candidate samples of Linux kernel vulnerabilities, dynamically reproduced 1724 vulnerabilities, and generated 948 local privilege escalation (LPE) exploits.
Why it matters: OpenAI disclosed a five-stage process and cost-benefit comparison of its automated vulnerability research pipeline, which can help understand the actual output of agents in security research.
OpenAI releases a security report, where its next-generation cybersecurity model identified 2308 potential security issues during a test targeting the Linux kernel, successfully reproduced 1724 vulnerabilities, and generated 948 Exploits for local privilege escalation
The Bluetooth SIG has adopted 'High Data Throughput' (HDT) technology as an optional feature for Bluetooth Core Specification 7.0, increasing the peak bit rate of low-power transmission from 2 Mbps to 7.5 Mbps, a speed increase of 275%. HDT enhances spectral efficiency, energy efficiency, and reliability through a new physical layer, with the first flagship scenario being high-end lossless wireless audio, with the first supporting devices expected to be released in early 2027.
OpenAI NewsDevRev uses GPT-6.1 Sol to drive computers and operators, balancing reasoning, speed, and cost to serve businesses.
OpenAI NewsFeatured6767OpenAI has built an autonomous pipeline to find, verify, and generate patches for 948 local privilege escalation vulnerabilities in the Linux kernel. The material only provides this result, not disclosing the specific methods of the pipeline, vulnerability details, or patch implementation status.
Why it matters: OpenAI uses an autonomous pipeline to locate and verify 948 local privilege escalation vulnerabilities in the Linux kernel, observable level of automation in the security research process.
TechCrunch · AISame story · 3 outlets6565Nothing CEO Carl Pei announced on X platform that the company's first 'AI-native product' will be released in December, but the specific form of the product remains unclear. The CEO described it as a 'non-screen, non-app productivity tool' and hinted that it is unrelated to AI wearable devices, sparking market speculation.
NVIDIA Technical Blog6565NVIDIA has developed the AI Factory Operations Agent to simplify operations within DGX SuperPOD, automatically correlating compute, network, hardware, and scheduling issues through a single interface to help SREs identify the causes of training performance degradation. The agent can process data from multiple dashboards, improving operational efficiency.
MarkTechPostSame story · 4 outletsRicursive AI has released Ripple-65B Preview, a 65B dense open-source model, scoring 80.5 on Terminal-Bench 2.1, becoming the highest-scoring open-source model in the United States. Its parameter size is 65B (active parameters), pre-training used 12T tokens, with a maximum context length of 1M, and is released under the Apache 2.0 license.
MarkTechPostSame story · 2 outletswebAI has released TwIL-8B, an 8.79B dense inference model fine-tuned based on IBM Granite 4.2 8B, achieving 98.2 on MATH-500 and 81.6 on LiveCodeBench†, using 38% fewer tokens than Qwen3.5-9B.
TechCrunch · AI6565OpenAI's Chief Global Affairs Officer Chris Lehane believes the 2028 presidential election could be a turning point for AI, urging policymakers to create policies that address the concerns of the American people about AI, rather than merely opposing AI and data centers. He proposed five policy priorities, covering worker protection and education, AI safety, and data center development. This stance marks a significant shift in OpenAI's policy agenda, moving away from last year's focus on accelerating AI development, reducing regulatory barriers, and expanding infrastructure.
TechCrunch · AI5050Uber stated in its third-quarter earnings call that during the busiest week of the Halloween period in October, there were only 4 Uber rides booked through an AI agent, and the AI agent also completed only 500 food orders through Uber Eats, which accounted for a very small proportion of the total 320 million trips and orders of Uber and Uber Eats for that week.
NewcomerSame story · 2 outlets6565Investors are comparing the funding trajectories of OpenAI and Anthropic, with the latter pulling away from the former, which is eager to tell a story of revival. Despite challenges in the AI industry, according to investors, Anthropic's IPO is still on track, with its stock having 'limitless demand' among long-term investors, and plans to complete it before Thanksgiving. Anthropic's more favorable stance towards the Democratic Party also helps its case.
Epoch AI6565Epoch Capabilities Index (ECI) update adopts Bayesian model methodology, providing more reliable estimates of model capability uncertainty and the probability of comparing capabilities between models. The new method is based on four-parameter logistic (4PL) item response theory (IRT), using Claude 3.5 Sonnet (ECI 130) and GPT-5 (ECI 150) as anchor models, and achieves cross-temporal consistency through linear transformations to fix parameter location and scale.
Tomasz TunguzThe upper limit of AI computing is expected to reach 80 gigawatts in 2027, with the annual growth rate doubling, but the cost per GPU hour will double within half a year to $12-13, with Anthropic's GPU hour cost rising by 82% in two quarters. The growth of GPU supply and price will double simultaneously, with physical reality becoming a bottleneck in demand, continuously pushing up the cost per GPU hour. Theoretically, AI revenue will increase from $140 billion at 38 gigawatts to nearly $500 billion at 43 gigawatts, but high costs may force model companies to adjust the allocation of GPU timing between training and inference, limiting the speed of innovation.
PyTorch Blog3737The call for proposals for PyTorch Day China 2026 ends on November 10 at 23:59 China Standard Time. The event is scheduled for December 12 in Beijing and is open to PyTorch developers, machine learning engineers, AI researchers, and open-source AI leaders from China and around the world. Proposals are welcome for talks and lightning talks covering five themes: model training and post-training, efficient inference and deployment, PyTorch core and hardware acceleration, physical AI, and PyTorch ecosystem and applications.
TechCrunch · AISame story · 4 outletsMeta has acquired Conductor, an AI programming startup that allows users to launch software projects by deploying multiple agents in parallel workspaces, Conductor's core product was launched last summer.
O'Reilly RadarCommon Fabric's CEO Alex Komoroske articulated the core concepts of his Resonant Computing Manifesto in a conversation with Tim O’Reilly. The manifesto proposes five principles: Private, Focused, Diverse, Adaptive, and Altruistic, aimed at creating technology experiences that nourish rather than consume users.
Ars Technica · AIFeatured8080In the U.S. midterm elections, 38 out of 44 measures restricting data centers in 10 states were passed, including multiple moratoriums and bans; Democrat Gina Hinojosa won the Texas gubernatorial election, planning to delay projects until legislation involving water, electricity, air, and property rights is passed.
Why it matters: The election results and industry statements create a contrast, allowing readers to judge the policy resistance faced by U.S. AI infrastructure expansion.
Gemini CLI releases1010Gemini CLI v0.67.0 released, adds the ability to bridge the IDE companion to container sandboxes, and fixes a bug where agents are repeatedly loaded when the workspace is the home directory.
The Verge · AI6565AT&T Connected Life Smart Security System becomes the first product to connect to Google Home Premium, enabling AI features like event summaries and video search through Gemini for Home, supporting natural language queries for activity logs. This integration is only available for users of the monthly $29.99 AT&T Professional Monitoring Plan, and adds a new feature for automatic arming/disarming based on user location.
Databricks BlogThe company is re-evaluating its reliance on a single cutting-edge model provider due to seven pressures: insufficient cost-effectiveness, frequent changes in model rankings, front-end model restrictions on legal work, changes in data retention policies, lack of visibility in model governance, model framework affecting performance, and the performance improvement of open-source weighted models.
AMD's Q4 revenue is expected to reach $17 billion, setting a record and surpassing Intel for the first time; surging sales of data center hardware offset declines in the client sector, with Epyc and Instinct product revenue doubling in the past two years.
The DecoderFeaturedSame story · 4 outlets7171ChatGPT now supports generating dynamic GIFs, this feature is based on OpenAI's new Images 2.6 model, and is being pushed to all ChatGPT and Codex users on web, desktop, and mobile platforms.
Why it matters: ChatGPT has added GIF generation capabilities, allowing readers to understand the progress of image models in cross-frame consistency and API variant divisions.
Codex releases00GitHub Copilot 0.168.0-alpha.2 released, this AI coding assistant is used for tasks, projects, and automation work, providing code review, quality analysis, vulnerability detection, and other functions.
MIT Technology Review · AITexas, the fastest-growing data center market in the US and the center for AI construction of large technology companies, may face a threat. The Democrats won majorities in the state Senate, Governor, and State Legislature in the midterm elections, with candidates frequently discussing data center issues
Gemini CLI releases1515Gemini CLI v0.68.0-preview.0 released, fixed 35 issues including terminal width change history re-rendering, case-insensitive isWithinRoot check on Windows, dot-included Gemini 3 model identification, and $-pattern corruption in LLM prompt templates. Added extension validation command and evaluation test.
