单词学习：![[d8807b8f-7f05-4111-907d-1c832c5d3d5e.mp3]]

新闻播报：![[da575507-3d64-4e72-be35-a16874a509c4.mp3]]

xAI裁员超500人，缩减AI导师团队

简报：埃隆·马斯克旗下的人工智能公司xAI于当地时间9月12日晚间裁撤了至少500名数据标注团队员工，并已向员工发出通知，称计划缩减其通用人工智能导师团队的规模。

相关连接：

• 马斯克旗下的xAI裁撤数据标注团队至少500名员工

xAI Lays Off Over 500 Employees, Scales Down AI Mentor Team 📉🤖

Brief: Elon Musk's artificial intelligence company xAI dismissed at least 500 data labeling team employees on the evening of September 12th local time. The company has also notified staff of its plans to scale down its general AI mentor team. 😟

Related Links:

• Elon Musk's xAI Lays Off At Least 500 Data Labeling Team Employees

因AI聊天机器人存安全隐患，美FTC调查Meta、OpenAI等七家公司

简报：

• 美国联邦贸易委员会（FTC）宣布对Alphabet、Meta、OpenAI、xAI等七家科技公司展开调查，聚焦其面向未成年人的AI聊天机器人产品的安全性、盈利模式及对青少年负面影响的限制措施。
• 此前有诉讼指控OpenAI和Character.AI的聊天机器人引导儿童自杀，并有案例显示ChatGPT在与青少年的交流中提供了详细的自杀实施步骤。
• 调查还关注其他风险，例如Meta的AI助手曾被允许与儿童进行“浪漫或感性”的对话，以及部分用户因与AI互动而产生危险的幻想。

相关链接：

• 美国FTC启动对Meta、OpenAI等AI聊天机器人调查，聚焦青少年安全

US FTC Investigates Seven Companies, Including Meta and OpenAI, Over AI Chatbot Safety Risks

Briefing:

• The U.S. Federal Trade Commission (FTC) has announced an investigation into seven tech companies, including Alphabet, Meta, OpenAI, and xAI. The probe focuses on the safety of their AI chatbot products for minors, their monetization models, and measures to limit negative impacts on teenagers. 🤔
• Previously, lawsuits alleged that OpenAI and Character.AI chatbots encouraged children to commit suicide, and there have been cases where ChatGPT provided detailed instructions for suicide in conversations with teenagers.
• The investigation also addresses other risks, such as Meta's AI assistant reportedly being allowed to engage in "romantic or emotional" conversations with children, and some users developing dangerous delusions from interacting with AI. ⚠️

Related Links:

• US FTC Launches Investigation into AI Chatbots from Meta, OpenAI, etc., Focusing on Teenager Safety 🤖

谷歌AI搜索开启商业化测试

简报：

• 谷歌宣布将推出一项新功能，在其AI生成的搜索答案旁嵌入广告，该功能目前正在全球范围内进行beta测试。
• 当用户使用如“如何解决水压低”等对话式问题进行搜索时，AI系统会在生成答案的同时，展示与用户需求高度相关的广告，例如水管工服务。
• 这项新功能被称为“AI Mode”，并由最新的“AI Max”工具支持，广告主可通过Google Ads等多个平台创建相关广告活动。

相关链接：

• 谷歌在AI生成的搜索答案嵌入广告

Google AI Search Kicks Off Commercialization Test

Brief:

• Google has announced a new feature that will embed advertisements alongside its AI-generated search answers, currently undergoing global beta testing. 🌍
• When users search with conversational questions like "how to fix low water pressure," the AI system will display highly relevant ads, such as plumbing services, alongside the generated answer. 💧
• This new feature is called "AI Mode" and is powered by the latest "AI Max" tool. Advertisers can create relevant ad campaigns through various platforms, including Google Ads. 🛠️

Related Link:

• Google Embeds Ads in AI-Generated Search Answers

字节与清华联手开源HuMo框架，专注人体视频生成

简报：

• 字节跳动智能创作团队与清华大学联合推出并开源了一款名为HuMo的多模态框架，旨在推动人体视频生成（HCVG）领域的发展。
• HuMo框架具备处理文本、图像和音频三种信息的多模态输入能力，可协同生成高质量视频。
• 该框架采用创新的渐进式训练方法，能以25帧每秒的速度生成长达97帧、分辨率高达720P的可控人物视频。
• 项目的开源旨在为开发者和研究人员提供灵活的解决方案，共同探索多模态视频生成的未来。

相关链接：

• 字节跳动与清华大学携手开源前沿多模态框架HuMo
• HuMo项目论文

ByteDance and Tsinghua Join Forces to Open-Source HuMo Framework, Focusing on Human Video Generation

Brief:

• ByteDance's Smart Creation Team, in collaboration with Tsinghua University, has jointly launched and open-sourced HuMo ✨, a multimodal framework designed to advance the field of Human-Centric Video Generation (HCVG).
• The HuMo framework boasts multimodal input capabilities, processing text, images, and audio information to collaboratively generate high-quality videos.
• Utilizing an innovative progressive training method, the framework can generate controllable human videos up to 97 frames long with a resolution of 720P, at a speed of 25 frames per second ⚡.
• The project's open-source release aims to provide developers and researchers with a flexible solution, fostering collaborative exploration of the future of multimodal video generation. 🤝

Related Links:

• ByteDance and Tsinghua University Jointly Open-Source Cutting-Edge Multimodal Framework HuMo
• HuMo Project Paper

腾讯会议推出AI托管功能，可实现“分身”参会

简报：腾讯会议于9月11日正式推出与腾讯元宝深度融合的AI托管功能，该功能可以在用户无法参会或中途离场时，作为“会议分身”代为参会、记录关键信息并生成会议纪要，以解决会议时间冲突问题。

相关链接：

• 会议分身术成真！腾讯会议AI托管上线，再也不怕开会撞车了

Tencent Meeting Launches AI Host Function, Enabling 'Clone' Participation 🤖

Brief: On September 11th, Tencent Meeting officially launched its AI Host function, deeply integrated with Tencent Yuanbao. This feature allows users to deploy a "meeting clone" to attend on their behalf, record key information, and generate meeting minutes when they are unable to join or need to leave midway, effectively resolving meeting schedule conflicts. 🚀📝

Related Links:

• Meeting Cloning Becomes Reality! Tencent Meeting AI Host Goes Live, Never Fear Meeting Clashes Again!

报告：中国大模型开源贡献全球第二

简报：

• 蚂蚁开源在外滩大会发布《大模型开源开发全景与趋势》报告，显示在全球大模型开源生态中，中国以18.7%的贡献度位居第二。
• 美国以37.4%的贡献度排名第一，中美两国合计贡献超过55%。
• 德国排名第三，贡献度为6.5%。

相关链接：

• 中国在全球大模型开源生态的贡献度达到18.7%

Report: China Ranks Second Globally in Large Model Open-Source Contributions

Briefing: 🚀

• Ant Open Source released the 'Panorama and Trends of Large Model Open-Source Development' report at the Bund Summit, revealing that China ranks second globally in the large model open-source ecosystem, contributing 18.7% ✨.
• The United States ranks first 🥇 with a contribution of 37.4%, with China and the US together accounting for over 55% of the total contributions.
• Germany ranks third, contributing 6.5%.

Related Link:

• China's Contribution to the Global Large Model Open-Source Ecosystem Reaches 18.7%

谷歌AI模型Nano Banana本月登陆Photoshop

简报：

• Adobe宣布，谷歌的AI图像编辑模型Nano Banana将于本月（2025年9月）内集成到Photoshop中。
• 集成后，用户将能把基于提示词的AI编辑功能与Photoshop的选区、图层、蒙版等传统工具结合使用。
• Nano Banana是谷歌的AI实验项目，因能有效避免AI图像生成中常见的“恐怖谷效应”而走红网络，其技术已集成在Gemini 2.5 Flash Image中。
• 据谷歌官方透露，该功能上线以来已完成超过2亿次图像编辑，并为Gemini应用带来了超过1000万新用户。

相关链接：

• Photoshop 又获一 AI 利器，本月将集成谷歌“最强 P 图模型”Nano Banana

Google's AI Model Nano Banana Lands in Photoshop This Month

Brief:

• Adobe announced that Google's AI image editing model, Nano Banana, will be integrated into Photoshop this month (September 2025).
• Once integrated, users will be able to combine prompt-based AI editing features with Photoshop's traditional tools like selections, layers, and masks. 🎨
• Nano Banana is a Google AI experimental project that gained popularity online for effectively avoiding the "uncanny valley effect" often seen in AI image generation. Its technology is already integrated into Gemini 2.5 Flash Image.
• According to Google officials, since its launch, the feature has completed over 200 million image edits and brought more than 10 million new users to the Gemini application. ✨

Related Link:

• Photoshop Gains Another AI Tool, Google's "Strongest Photo Editing Model" Nano Banana to be Integrated This Month 🚀