Article summary — original

Anthropic Adds Invisible Watermarks to Claude as xAI Debuts Grok Bot Agents and River AI Raises $1.1B Anthropic has begun embedding invisible watermarks in Claude outputs to comply with global transparency regulations, while xAI and Cursor rolled out Grok Bot and former xAI co-founder Igor Babuschkin secured $1.1B for personal AI startup River AI. Anthropic is rolling out invisible watermarks for text, code, and files generated by Claude to comply with EU AI Act transparency rules globally. SpaceXAI and Cursor launched a beta of Grok Bot, an iMessage-style interface managing autonomous AI agent teammates on dedicated computers. River AI, founded two months ago by former xAI co-founder Igor Babuschkin, raised $1.1B to build privately owned, open-source AI models. OpenAI COO Brad Lightcap announced his departure to launch a new venture. Nvidia released Nemotron 3.5 Lightning, and LTX launched its open-weights world model LTX-2.5. Anthropic embeds invisible watermarks across Claude outputs Anthropic detailed plans to embed invisible watermarks into Claude-generated text, code, and files, aligning its global platform with EU AI Act transparency mandates. New models shipping after August 2 include the built-in tracking that persists through copy-pasting, while older models will be retrofitted. Files will adopt the C2PA provenance standard, and Anthropic plans to release detection tools clarifying that flagged content was processed by Claude rather than necessarily authored entirely by it. SpaceXAI and Cursor introduce Grok Bot agent teammates Grok Bot is a new beta application creating an iMessage-style team chat populated by autonomous AI agents that operate on dedicated virtual computers. These bots can coordinate independently, create specialized sub-agents, learn workflows, and access apps and websites 24/7 without requiring user hardware to stay open. The beta is currently accessible across iPhone, Mac, Windows, and Linux desktops for SuperGrok Heavy and top-tier Cursor users. River AI raises $1.1B for privately controlled open AI Two months after launching, startup River AI secured $1.1B to develop open-source AI systems designed to run on users' private hardware rather than corporate cloud infrastructure. Founder Igor Babuschkin, an alumnus of OpenAI, Tesla, Google, and co-founder of xAI, stated the goal is to prevent frontier corporate labs from monopolizing powerful models. The company launched a live API claimed to fine-tune open-weight models in minutes. Industry briefs and tool updates OpenAI Chief Operating Officer Brad Lightcap is departing to start a new company focused on AI's next phase. Anthropic has signaled to investors an increased focus on biology and healthcare initiatives. Meanwhile, Nvidia released the compact Nemotron 3.5 Lightning for agent workflows while developing Nemotron 4, LTX launched the open-weights LTX-2.5 world model, and OpenAI brought ChatGPT Desktop to Linux.

Article summary — zh

Anthropic 为 Claude 输出内容嵌入隐形水印,多项 AI Agent 与开源项目同步更新 本期 AI 简讯聚焦 Anthropic 实施欧盟 AI 法案透明度规则,在其模型输出中嵌入隐形水印;同时涵盖 xAI 与 Cursor 合作推出的 Grok Bot 协作系统、初创公司 River AI 完成 11 亿美元融资等业界最新动态。 Anthropic 计划在 Claude 生成的文本、代码及文件中植入隐形水印以落实透明度要求 xAI 与 Cursor 联合推出具备独立虚拟设备并支持自主协作的 Grok Bot 代理应用 xAI 前联合创始人 Igor Babuschkin 成立的开源 AI 初创企业 River AI 获得 11 亿美元融资 OpenAI 首席运营官 Brad Lightcap 宣布离职开展新项目 LTX 发布视频世界模型 LTX-2.5,英伟达推出面向智能体任务的 Nemotron 3.5 Lightning Anthropic 推出隐形水印以标记 Claude 生成内容 Anthropic 发布支持页面说明其隐形水印实施方案,覆盖文本、代码和文件输出,以全面践行欧盟《人工智能法案》的透明度要求。新模型生成的内容在复制到站外时仍会保留隐藏标识,文件则接入 C2PA 行业来源标准。今年 8 月 2 日后发布的模型已内置该功能,旧模型将逐步改装。官方还将推出专属检测工具,并澄清标记仅代表内容由 Claude 处理,并不代表其完全由 AI 原创。 xAI 推出协同代理 Grok Bot,River AI 获 11 亿美元融资 xAI 携手 Cursor 推出 Grok Bot 测试版,该应用以类 iMessage 的群聊形式运行,智能体拥有专属虚拟计算设备,可跨应用自主执行流程并支持 24/7 全天候协作。与此同时,前 xAI 联合创始人 Igor Babuschkin 创办仅两个月的 River AI 筹集了 11 亿美元,该团队主打由用户私人硬件托管并全权控制的开源 AI 助手,以对抗封闭式前沿实验室的中心化掌控。 行业快讯与模型发布 OpenAI 首席运营官 Brad Lightcap 宣布离职创业;Anthropic 向投资者透露将把医疗和生物学领域作为发展重点以改善公众情绪。在模型发布方面,视频生成公司 LTX 推出了开源权重世界模型 LTX-2.5,英伟达则发布了小体量高效模型 Nemotron 3.5 Lightning 并透露正筹备 Nemotron 4;此外 ChatGPT 桌面客户端已拓展支持 Linux 平台。

Saved original article

[Read Online]({{live_url}}) <sup>|</sup> [Sign Up](https://www.therundown.ai/subscribe?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) <sup>|</sup> [Advertise](https://therundownai.typeform.com/to/kraZ1TSO?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) **Good morning, {{ first\_name | AI enthusiasts }}, and welcome to the 5,596 new readers who joined us yesterday.** That memo you just worked with Claude to write? It might soon carry an invisible signature that says so. Anthropic just quietly detailed the rollout of invisible watermarks that label everything its Claude models touch, and, like most features forced onto the world, a lot of users aren’t happy about it. **Reminder:** *Our next live workshop is today at 2 PM EST. Join University Instructor Nate Grehek for “Your AI Reset”, a beginner catch-up on the latest terms, tools, and setups to get better results in your work.* [RSVP here](https://app.therundown.ai/live/your-ai-reset-workshop?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude)*.* **In today’s AI rundown:** Anthropic adds AI watermarks to Claude outputs Grok Bot is the latest AI agent teammate build Use ChatGPT to build a custom Mac shortcut system xAI co-founder’s River banks $1.1B for personal AI **LATEST DEVELOPMENTS** Image source: Images 2.0 / The Rundown **The Rundown:** Anthropic just [published](https://support.claude.com/en/articles/16266773-how-claude-marks-ai-generated-content?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) a support page detailing plans to embed invisible watermarks in text, code, and file outputs from Claude, putting the EU AI Act’s transparency rules into practice globally across its user base. **The details:** Text and code generated by new Claude models will carry a hidden watermark that carries through when copied and pasted outside the platform. Files will get the same C2PA provenance standard label that is already used across the industry to mark AI-generated media. Older Claude models will need to be ‘retrofitted’ to include the labelling, with models shipping after Aug. 2 having the tool built in. Anthropic is also planning to [release](https://x.com/trq212/status/2087258090169414008?s=20&utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) its own detection tools and noted that a mark shows content was “processed by Claude”, not necessarily fully authored. **Why it matters:** The move is a polarizing one, with takes ranging from positive to dystopian and overreaching. Anthropic isn’t alone in signing the AI Act, with OpenAI, Google, Meta, and others (notably absent: xAI) likely needing to navigate the issue as well. For the negative camp, the case for private and open models continues to grow. **The Rundown:** Search results go stale, so teams bolt on an API, add a scraper, wire up a parser, and hope it holds. Yeah, it works in the demo, but then it breaks under production load six months later. This guide from [You.com](https://you.com/resources/ydc-apis?utm_campaign=28104480-The%20Rundown%20Q3&utm_source=external-newsletter&utm_medium=email&utm_term=rundown_primary_715&utm_content=rundown_primary_715) explains why treating research as simple retrieval is the wrong frame, and what a purpose-built stack looks like instead. **What you'll learn:** Why general-purpose search hits 93% SimpleQA accuracy but still fails high-stakes workflows The difference between an API that returns results and one that answers questions How teams reclaim 6–12 months of time with a single endpoint What finance, legal, and compliance workflows need beyond web search Stop treating research as retrieval. [Download the guide](https://you.com/resources/ydc-apis?utm_campaign=28104480-The%20Rundown%20Q3&utm_source=external-newsletter&utm_medium=email&utm_term=rundown_primary_715&utm_content=rundown_primary_715). Image source: xAI **The Rundown:** SpaceXAI and Cursor just [rolled out](https://x.ai/news/introducing-grok-bot?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) Grok Bot, a new beta app that turns Grok into an iMessage-style chat of teammates that have their own dedicated computers, work autonomously, and can access everyday apps just like a real user. Bots are able to coordinate with each other independently or in group chats, and learn from workflows and improve over time. New specialist agents can be generated by Bots inside a job or receive handoffs from another worker, with work continuing 24/7 without a laptop open. Bots can also access websites and apps on a user’s behalf and observe and create workflows to automate tasks. The beta covers iPhone and Mac, Windows, and Linux desktops for SuperGrok Heavy and top Cursor tiers, with Elon Musk [saying](https://x.com/elonmusk/status/2087233507370147920?s=20&utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) access will expand soon. **Why it matters:** The group chat is becoming a top form factor for running teams of agents. [OpenClaw](https://app.therundown.ai/courses/launch-your-openclaw-from-scratch?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) and Hermes already live in messaging apps, and Jack Dorsey's [Buzz](https://app.therundown.ai/guides/build-project-room-humans-ai-agents-buzz-tutorial?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) built a Slack rival where agents are channel members. Grok Bot feels like a middle-ground version built for the masses (if the price comes down significantly). **The Rundown:** In this guide, you will learn how to use ChatGPT Work to set up time-saving keyboard shortcuts and automations for you on your Mac, with no installations needed to get started. **Step-by-step:** Open ChatGPT Work and say: “Based on what you know about me and my workflow, suggest 10 time-saving hotkeys we can set up in Apple Shortcuts” Keep the shortcuts that will save you the most time. Say: “Keep the highest-frequency shortcuts and avoid macOS or app conflicts” Enable Computer Use and paste the approved list, saying: “Set these up in Apple Shortcuts. Do not change anything outside this list. Test each shortcut” Let ChatGPT build them, then press each key to test it out. Also ask for “a one-page cheat sheet of every shortcut, where it lives, and what is left to do” **Pro tip**: Ask ChatGPT Work for the top 10 time-saving automations it would set up for you. It can create them in Automator and pull them directly into Shortcuts. **The Rundown:** Pave turns the data your team already has into real, working software, not a prototype or a rebuild. Upload a spreadsheet or describe it, and get an app that's structured, governed, and built to run. **With Pave you can**: See real-time data across your operation Get workflows that match how your team works Give your team shared visibility they can rely on Image source: River AI / The Rundown **The Rundown:** Former xAI co-founder Igor Babuschkin just [raised](https://x.com/ibab/status/2087145290671468569?s=20&utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) $1.1B for River AI, his two-month-old startup focused on building open-source AI that is trained, tuned, and controlled by the individual instead of a major corporation. Before xAI, which he left a year ago after helping co-found the startup with Elon Musk in 2023, Babuschkin worked at OpenAI, Tesla, and Google. River claims its live API, released at launch, turns open-weight models into ones a business actually can own, cutting training to just minutes. Babuschkin [told](https://www.nytimes.com/2026/08/11/technology/igor-babuschkin-xai-river-ai.html?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) the NYT the goal is a highly customizable assistant that follows across devices while running on the user’s own private hardware. He also said, “We don’t want these AI companies to rule the world and control this superpowerful technology,” hitting at closed U.S. frontier labs. **Why it matters:** It’s another eye-popping raise for a company without much to show yet, but Babuschkin’s pedigree is worth betting on. With questions about access (Fable/Mythos ban) and forced watermarks (above) putting privacy and control top of mind in the industry, River is pulling at a string with a whole lot of interest. **QUICK HITS** Today’s workflow comes from reader **Alastair Cross**: “My mum’s 80th birthday is next week, and my dad, sister, and I wanted to create a photobook of her life. My dad scanned hundreds of photos from over the years and sent them to me to clean up. Because we were on a tight deadline, I didn’t have time to open and edit each image individually. I asked ChatGPT to build a tool that accepts a folder of scanned images in various formats, including single photos, overlapping photos, and photos cut off by the scanner. The tool processed each, cropped and straightened the photos, and provided a UI where I could review and make adjustments. I then configured the tool to output the images to Google Drive and created a spreadsheet containing thumbnails of every image, along with a rating system for photos we wanted to include in the final book.” *See Alastair’s workflow here. How do you use AI? Tell us for a chance to be featured.* 🤖 [Grok Bot](https://www.rundown.ai/tools/grok-bot?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) - xAI's always-on agent teammates with their cloud computer 🎬 [LTX-2.5](https://www.rundown.ai/tools/ltx-2-5?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) - LTX's open-world model for video, real-time avatars, robotics ⚡ [Nemotron 3.5 Lightning](https://www.rundown.ai/tools/nemotron-3-5-lightning?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) - Nvidia's small, open model for agentic work 💻 [ChatGPT Desktop](https://x.com/OpenAI/status/2087231350134980830?s=20&utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) - App for ChatGPT, Work, and Codex, now on Linux **OpenAI COO Brad Lightcap** is [leaving](https://x.com/TheRundownAI/status/2087223385772204112?s=20&utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) to “start something new,” saying there are “a few important new things the world will need to get right” as AI enters its next phase. **Anthropic** reportedly [told](https://t.co/Z5InpzFupv?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) investors it plans to prioritize AI in healthcare and biology to help “mitigate some of the negative sentiment around AI”. **AI video company LTX** [released](https://ltx.io/newsroom/introducing-ltx-2-5?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) LTX-2.5, an open-weights world model that tops rivals like Gemini Omni Flash on internal tests for video generation speed and quality. **Nvidia** [released](https://developer.nvidia.com/blog/nvidia-nemotron-3-5-lightning-delivers-fast-accurate-specialized-task-execution-for-long-running-agents/?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) a small, fast Nemotron 3.5 Lightning model and [anticipates](http://theinformation.com/articles/nvidia-trying-develop-worlds-best-open-source-ai-models?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) its Nemotron 4 to be “on par with the best open-source AI models in the world.” Read our last AI newsletter: [Meta returns to its open-source roots](https://www.therundown.ai/p/meta-returns-to-its-open-source-roots?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) Read our last Tech newsletter: [Apple’s almost-glass iPhone](https://tech.therundown.ai/p/apple-almost-glass-iphone?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) Read our last Robotics newsletter: [Uber’s $10B answer to Waymo](https://robotnews.therundown.ai/p/uber-10b-answer-to-waymo?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) Today’s AI tool guide: [Use ChatGPT to build a custom Mac shortcut system](https://app.therundown.ai/guides/how-to-build-a-custom-mac-shortcut-system-for-faster-workflows?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) RSVP to next workshop today at 2 PM EST: [Your AI reset](https://app.therundown.ai/live/your-ai-reset-workshop?utm_source=www.therundown.ai&utm_medium=newsletter&utm_campaign=anthropic-slips-an-invisible-signature-into-claude) See you soon, *Rowan, Zach, Shubham, Jennifer, and Nate — the humans behind The Rundown*