![[69d07402-653c-446d-b9e3-ed865e7df526.mp3]]

AI代码生成独角兽Cognition AI以98亿美元估值完成新一轮融资

简报：

• 据报道，人工智能独角兽公司Cognition AI在新一轮融资中，以98亿美元的估值筹集了约5亿美元资金。
• 本轮融资获得的资金将主要用于推进公司的人工智能代码生成业务。

相关链接：

• 独角兽Cognition AI按98亿美元估值实施最新一轮融资

AI Code Generation Unicorn Cognition AI Completes New Funding Round at $9.8 Billion Valuation 🦄

Brief:

• Reportedly, AI unicorn Cognition AI has raised approximately $500 million in a new funding round, valuing the company at$9.8 billion. 💰
• The capital secured from this round will primarily be allocated to further advance the company's AI code generation operations. 💻

Related Link:

• Unicorn Cognition AI Conducts Latest Funding Round at $9.8 Billion Valuation

Midjourney开放视频生成，标准用户可制作高清短片

简报：

• 人工智能公司Midjourney发布了其首个视频模型（V1），允许用户将Midjourney生成或自己上传的图片制作成动画短片。
• 此次升级后，标准订阅用户可以生成高清视频，并支持批量制作。
• 新功能还包括优化的视频审核功能以确保内容合规，以及一个独立的Moodboards页面用于管理创意灵感。

相关链接：

• Midjourney 重磅升级！标准用户也能轻松生成高清视频！
• Midjourney adds AI video generation - Engadget
• Midjourney, Amid Disney/NBCU Lawsuit, Launches Video ... - Variety

Midjourney Unlocks Video Generation, Standard Users Can Now Create HD Short Films 🎬

Brief:

• AI company Midjourney has released its first video model (V1), allowing users to animate images generated by Midjourney or their own uploaded pictures into short video clips. ✨
• Following this upgrade, standard subscribers can generate high-definition videos and also benefit from batch production support.
• New features also include optimized video moderation to ensure content compliance ✅, and a dedicated Moodboards page for managing creative inspirations.

Related Links:

• Midjourney 重磅升级！标准用户也能轻松生成高清视频！
• Midjourney adds AI video generation - Engadget
• Midjourney, Amid Disney/NBCU Lawsuit, Launches Video ... - Variety

谷歌相册将上线新功能，可检测图片是否由AI生成或编辑

简报：

• 谷歌相册（Google Photos）将推出名为“How Was This Made”的新功能，旨在帮助用户识别图片和视频的真实性。
• 该功能通过读取嵌入媒体元数据中的“内容凭证”（Content Credentials），可显示内容是相机自然拍摄、经AI或非AI工具编辑，还是完全由AI生成。
• 根据对谷歌相册7.41版APK安装包的代码分析，用户可在图片详情中查看如“由AI创建的媒体”、“使用AI工具编辑”等具体信息。
• 此更新旨在弥补用户对AI技术的信任鸿沟，提升在深度伪造内容日益普遍环境下的数字内容透明度。

相关链接：

• https://www.aibase.com/zh/news/20524
• https://www.ithome.com/0/875/408.htm

Google Photos to Launch New Feature: Detecting AI-Generated or Edited Images

Brief:

• Google Photos is set to introduce a new feature called “How Was This Made,” designed to help users identify the authenticity of images and videos. 📸
• This feature works by reading 'Content Credentials' embedded in media metadata, indicating whether content was naturally captured by a camera, edited by AI or non-AI tools, or entirely generated by AI. 🕵️‍♀️
• According to an analysis of the Google Photos version 7.41 APK package code, users will be able to view specific information in image details, such as 'Media Created by AI' or 'Edited with AI Tools'.
• This update aims to bridge the trust gap users have with AI technology and enhance digital content transparency in an environment where deepfake content is increasingly prevalent. ✨

Related Links:

• https://www.aibase.com/zh/news/20524
• https://www.ithome.com/0/875/408.htm

Gemini赋能谷歌办公应用，Slides和Vids获AI图像编辑能力

简报：

• 谷歌宣布为在线演示文稿（Google Slides）和视频创作工具（Google Vids）推出两项基于Gemini模型的全新AI图像编辑功能。
• 新功能允许用户通过简单的文本提示替换图片背景，或将小型竖屏图片扩展为大型横屏横幅而不会扭曲原始图像。
• 该更新已于2025年8月14日面向指定的Google Workspace订阅用户及拥有相关AI插件的用户全面推出。

相关链接：

• https://www.aibase.com/zh/news/20499

Gemini Empowers Google Workspace Apps: Slides & Vids Gain AI Image Editing Capabilities ✨

Brief:

• Google has announced the launch of two new AI image editing features, powered by the Gemini model, for its online presentation tool (Google Slides) and video creation tool (Google Vids). 🖼️
• These new functions allow users to replace image backgrounds using simple text prompts, or expand small portrait-oriented images into large landscape banners without distorting the original image.
• The update was fully launched on August 14, 2025, for designated Google Workspace subscribers and users possessing the relevant AI plugins. 🚀

Related Link:

• https://www.aibase.com/zh/news/20499

智元机器人开源首个机器人世界模型平台

简报：

• 智元机器人发布了面向真实世界机器人操控的统一世界模型开源平台——Genie Envisioner（GE）。
• 该平台将未来帧预测、策略学习与仿真评估整合进以视频生成为核心的闭环架构，实现了机器人从“看”到“想”再到“动”的端到端推理与执行。
• GE基于约3000小时的真实机器人操控视频数据进行训练，在跨平台泛化和长时序任务执行上表现出显著优势，仅需少量数据即可实现平台迁移。
• 智元机器人宣布将开源该平台的全部代码、预训练模型和评测工具。

相关链接：

• Project page
• Arxiv
• Github
• IT之家报道

Agibot Robotics Open-Sources First Robot World Model Platform 🤖

Brief:

• Agibot Robotics has released Genie Envisioner (GE), a unified open-source world model platform designed for real-world robot manipulation.
• This platform integrates future frame prediction, policy learning, and simulation evaluation into a video generation-centric closed-loop architecture, enabling end-to-end reasoning and execution for robots, from "seeing" to "thinking" to "acting."
• GE was trained on approximately 3000 hours of real robot manipulation video data, demonstrating significant advantages in cross-platform generalization and long-horizon task execution, achieving platform migration with only a small amount of data. 🚀
• Agibot Robotics announced that it will open-source all code, pre-trained models, and evaluation tools for this platform. ✨

Related Links:

• Project page
• Arxiv
• Github
• IT Home Report

全球首届人形机器人运动会开赛，真实场景应用成焦点

简报：

• 全球首届世界人形机器人运动会于8月14日至17日在北京举行，共计280支队伍参赛。
• 参赛方包括宇树科技、星海图等国内企业，清华、北大等高校，以及来自美国、德国等15个国家的国际队伍。
• 本届赛事最大看点是场景赛，旨在考验机器人在酒店清洁、工厂物料搬运、医院药品分拣等真实场景下的综合能力。
• 除场景赛外，赛事还涵盖田径（百米和半程马拉松）、足球、搏击、舞蹈等传统竞技项目。

相关链接：

• 全球首届世界人形机器人运动会8月启幕:280支队伍角逐，场景赛成最大看点

First-Ever Global Humanoid Robot Games Kick Off, Real-World Applications Take Center Stage

Briefing:

• The First-Ever World Humanoid Robot Games were held in Beijing from August 14th to 17th, with a total of 280 teams participating. 🤖
• Participants included domestic enterprises such as Unitree Robotics and Xinghaitu, universities like Tsinghua and Peking University, as well as international teams from 15 countries including the United States and Germany.
• The biggest highlight of these Games was the scenario-based competitions, designed to test robots' comprehensive capabilities in real-world settings such as hotel cleaning, factory material handling, and hospital medicine sorting. 🏥
• In addition to the scenario competitions, the Games also featured traditional athletic events including track and field (100m sprint and half-marathon), football, combat, and dance. 🤸

Related Link:

• 全球首届世界人形机器人运动会8月启幕:280支队伍角逐，场景赛成最大看点

HTC发布AI眼镜VIVE Eagle，主打语音交互与多模型AI助理

简报：

• HTC于2025年8月14日发布其首款AI智能眼镜VIVE Eagle，需配合智能手机使用，机身含镜片重约48.8克。
• 该眼镜搭载高通骁龙AR1 Gen 1芯片，配备1200万像素摄像头，支持IP54防尘防水。
• AI功能方面，可调用谷歌Gemini和OpenAI GPT等大型语言模型，支持多语言即拍即译，并且是首款支持繁体中文语音控制的智能眼镜。
• 续航方面，内置235mAh电池，最长可支持4.5小时音乐播放或超过3小时的持续通话。
• VIVE Eagle将于9月1日在中国台湾地区率先发售，定价为15600新台币（约合3739元人民币）。

相关链接：

• HTC 发布首款 AI 智能眼镜 VIVE Eagle：骁龙 AR1 Gen 1 芯片，48.8g
• 台灣製造全球首發！HTC推AI眼鏡VIVE Eagle　功能、價格、上市日一次看

HTC Launches AI Glasses VIVE Eagle, Featuring Voice Interaction and Multi-Model AI Assistant 👍

Briefing:

• HTC unveiled its first AI smart glasses, VIVE Eagle, on August 14, 2025. These glasses require a smartphone for operation and weigh approximately 48.8 grams, including the lenses.
• The glasses are powered by the Qualcomm Snapdragon AR1 Gen 1 chip, feature a 12-megapixel camera, and boast IP54 dust and water resistance.
• For AI functionalities, it can access large language models like Google Gemini and OpenAI GPT. It supports real-time multi-language instant translation via camera and is the first smart glasses to support Traditional Chinese voice control. 🤯
• In terms of battery life, the VIVE Eagle includes a 235mAh battery, supporting up to 4.5 hours of music playback or over 3 hours of continuous calls.
• The VIVE Eagle will first go on sale in Taiwan on September 1, priced at NT$15,600 (approximately RMB 3,739). 🚀

Related Links:

• HTC 发布首款 AI 智能眼镜 VIVE Eagle：骁龙 AR1 Gen 1 芯片，48.8g
• 台灣製造全球首發！HTC推AI眼鏡VIVE Eagle　功能、價格、上市日一次看

美对华AI芯片销售附带关税，台积电或成赢家

简报：美国政府批准辉达（Nvidia）与超微（AMD）在缴纳15%关税的前提下，恢复向中国销售特制AI芯片，此举虽压缩了两家芯片巨头的利润空间，但因中国市场需求强劲，辉达已向台积电追加H20晶片订单，使台积电有望成为此次政策变动的主要受益者。

US Imposes Tariffs on AI Chip Sales to China, TSMC Set to Gain 💰

Brief: The U.S. government has approved Nvidia and AMD to resume selling specialized AI chips to China, subject to a 15% tariff. While this move compresses the profit margins of the two chip giants, strong demand in the Chinese market has led Nvidia to place additional orders for H20 chips with TSMC, making TSMC a likely primary beneficiary of this policy change. 📈🤖