Find tools, news, and tutorials
One place to explore all published AI content.
“AI”808 results in this category
AI news
808
Product Updates09/17/2026Anthropic Claude redesigned: Chat and Cowork interfaces merged, with PPT creation addedAnthropic has merged Claude’s chat interface with the Cowork frontend, so users no longer need to choose between tabs. It also adds presentation and Docs features, with support for exporting to PDF / PPT, collaborative editing, and continuing work on mobile. #Claude##Anthropic##AI Assistant#
Industry Watch09/17/2026Reports Say Anthropic to Make Its First Compute Infrastructure Move in Australia by Leasing a 2.16 GW Data Center CampusAustralia, with its vast open spaces, sparse population, and abundant renewable resources, is an ideal location for compute infrastructure. OpenAI also signed an agreement last year to build a data center in Australia.
Models & Technology09/13/2026Ant Lingbo Open-Sources Three LingBot-World 2.0 World Models: Small’s 1.3B Parameters Designed for Consumer Single-GPU SystemsAnt Lingbo Technology has further open-sourced three models: LingBot-World 2.0 Small (1.3B), LingBot-World 2.0 Bidirectional, and LingBot-World 2.0 Causal Pretrain.
Industry Watch09/12/2026Yu Minhong: Artificial Intelligence May Worsen the Wealth GapHe said, “This may be similar to the development and changes brought about by the Industrial Revolution, but it is more intense because it creates another phenomenon: artificial intelligence is concentrating humanity’s wealth in the hands of a very small number of people at an extremely rapid pace.”
Industry Watch09/11/2026Report: U.S. Department of Defense Considers Providing Neocloud Company Fluidstack with About $5 Billion LoanFluidstack will use the loan to strengthen U.S. supply chains and manufacturing capabilities for data center-related components, rather than directly investing in new artificial intelligence facilities.
Models & Technology09/10/2026NASA and IBM Jointly Release an Open-Source Foundation Model for Lunar Research, Capable of Identifying Lunar Ice Regions and CratersNASA and IBM have jointly launched an open-source lunar foundation model that reduces ice-region identification errors by 23% and improves performance by 19% using just half the training data. #LunarAI# They have also released the world’s first open-source dataset of its kind, containing more than 2 million data points to support scientific research.
Models & Technology09/10/2026DeepMind Engineer Rebuts “Google Can’t Build Frontier Models”: Gemini 3.8 Cyber Outperforms Mythos in Multiple TestsAn AI researcher claimed Google had lost its edge in frontier models. A DeepMind engineer rebutted the claim, saying Gemini 3.8 Cyber has surpassed Anthropic’s Mythos in cybersecurity and is widely used by internal teams. #AIRace# What do you think of this war of words?
Industry Watch09/09/2026IDC: Global Wrist-Worn Device Market Fell 4.3% Year Over Year in Q2 2026; Huawei, Xiaomi, and Apple Ranked Among the Top Three by ShareIDC released its report on the global wrist-worn device market in the second quarter of 2026, analyzing the market downturn, two key areas of exploration—on-device AI and screenless bands—as well as the performance of leading manufacturers, and offering an outlook for future development.
Models & Technology09/09/2026Huawei Pura X View and Mate XT 2 Extraordinary Master smartphones reportedly support downloading and deploying local large models on-deviceAccording to blogger @超维界, the Huawei Pura X View and Mate XT 2 Extraordinary Master smartphones support downloading and deploying local large models on-device. Users can choose to download either a multimodal enhanced large model (approximately 6 GB) or a multimodal mixture-of-experts large model (approximately 15 GB). The blogger said both models run on-device and can be used without a network connection. Huawei on-device AI large models#
Models & Technology09/09/2026The First Native Multimodal Model in the Ant Ling Series, Ling-3.0-flash-VL, Released as Open Source with a Visual Feedback LoopThe model uses a MoE architecture, supports multimodal input, and introduces a visual feedback loop. It performs well across multiple benchmarks and is now available for free online trials. Related versions have been released on open-source platforms. #AntLingLargeModel#
Models & Technology09/08/2026Zuckerberg Advances “Personal Superintelligence”: Meta Launches Its First Personal Agent, Muse, with Mixed Results in Internal TestingMuse can autonomously help users complete various tasks. It is initially available only in the United States, with free and paid subscription plans, but internal testing has revealed security vulnerabilities and operational failures. #MetaLaunchesPersonalAgentMuse#
Product Updates09/08/2026Google DeepMind Launches AlphaGenome Atlas with Predictions for All 9 Billion Single-Nucleotide Variants in the Human GenomeGoogle DeepMind has launched AlphaGenome Atlas, which includes functional predictions for all 9 billion possible single-nucleotide variants in the human genome. The atlas is freely available to academia and has already supported advances in research on rare diseases and other areas.
Robotics09/08/2026Faraday Future Robots Set New Monthly Sales Record of 158 Units in August, Cumulative Sales Surpass 552 UnitsJia Yueting released a video this morning announcing that sales and shipments of Faraday Future EAI robot units reached 158, setting a new monthly record. Jia Yueting said that since deliveries began in February, cumulative robot sales and shipments had reached 552 units by the end of August. This provides strong support for winning the Q3 robot commercialization campaign and pursuing the full-year target of 2000 units. #FaradayFuture# #EAIrobots#
Product Updates09/07/2026iRobot Launches Roomba Duo Mother-and-Child Cleaning RobotThe product consists of a dual-function vacuum-and-mop main unit and a compact companion unit. The two can communicate in real time and use AI to coordinate cleaning tasks.
Robotics09/06/2026Wujie Power CEO Zhang Yufeng Predicts Embodied Intelligence Could Match Adults' Labor Capabilities in 5 YearsWujie Power CEO Zhang Yufeng said at the Yabuli Forum Summer Summit that current embodied intelligence is comparable to autonomous driving in 2019, and predicted that its capabilities will reach the level of adults who can “think and work” in 5 years. He also believes humanoid robots could enter homes within 2–3 years.##Embodied intelligence##
Industry Watch09/05/2026Nearly 2 billion yuan in initial credit: Beijing E-Town first in the city to launch Token-based computing power loansBeijing E-Town has launched the “Yiqi Token Loan,” with nearly 2 billion yuan in initial credit extended to companies in the artificial intelligence industry chain. The product uses computing-power Token output and the value of computing-power service contracts as core lending criteria, overcoming the limitations of traditional collateral and providing a new financing channel for technology companies with high R&D investment. #TokenComputingPowerLoan#Product Updates09/03/2026Anthropic Upgrades Claude Code / Cowork to Control Users' Macs in the BackgroundAnthropic has announced an upgrade to Claude's computer control capabilities. It can now run in the background on macOS while handling actions such as clicks and text input without interrupting users' other work. The feature supports the Cowork and Code products and is available only to Pro and Max subscribers. #Claude# #AI Assistant#Product Updates09/02/2026Doubao Work Update: Supports Parallel Multi-Agent Tasks and Mac Computer ControlDoubao Work has received two major updates: first, the “Parallel Multi-Agent” feature can trigger multiple sub-agents to work together simultaneously, greatly improving efficiency; second, the Mac version now supports “Computer Control,” allowing it to understand and operate local interfaces without an API. #AIOffice# This marks another major move by ByteDance in AI-powered office applications.