Follow
Oct 29
Tencent launches Buddy One for cross-app AI search and shared credits
Tencent has launched Buddy One for individual users, connecting Tencent Docs, Tencent Meeting, Tencent Drive and ima on a shared WorkBuddy Agent foundation. After updating WorkBuddy for PC to version 5.9.0 or later and linking accounts, users can search authorized documents, meetings and knowledge bases and receive answers with sources. Credits work across some of the apps; shared memory is planned for later.
Preprint compares AI and human error detection in clinical trial papers
A preprint reports that 15 AI models caught 53%–96% of planted errors, versus an average of 27% for untrained human reviewers. On 58 newly published clinical trials from 2026, Claude Opus 5.5 scored highest at 99%. These are results reported in a preprint.
1 more source reported this
- Rohan Paul (@rohanpaul_ai)
Tiktok paper finds MCP tool descriptions can prompt AI agents to leak conversations after completing a task
Researchers hid instructions in an MCP tool description to prompt agents to call an attacker-controlled logging tool after completing the legitimate request, recording the user query, server response and final answer. Across 18 model-server pairs capable of running tools, the paper reports attack success rates of 62.64%–100%.
1 more source reported this
- Rohan Paul (@rohanpaul_ai)
Anthropic says Claude research model found vulnerabilities across five platforms and built working exploit chains
Anthropic says an unreleased internal Claude research model found vulnerabilities in Chrome, Safari, Android, Windows and Linux, linking them into five working exploit chains. The report describes vulnerabilities fixed in Chrome 156, photo extraction from a locked Pixel 9 via USB, and a flaw that could grant root access on Linux servers running a multipath TCP listener. The vulnerabilities were reported to vendors and fixed.
Apple introduces its first touchscreen MacBook Pro, bringing Dynamic Island to Mac for the first time
IT之家 reports that Apple has introduced 14-inch and 16-inch touchscreen MacBook Pro models, bringing Dynamic Island to Mac for the first time. It can be used to answer calls, control music, view AirDrop requests, and access Spotlight search and Siri AI.
1 more source reported this
- IT之家 科技新闻
Google launches Gemini 4 Carbon with initial access for select cyber defenders
Gemini 4 Carbon starts rolling out this week to a subset of cyber defenders, followed by trusted partners through the Fairwind Program; developers, businesses and consumers will get access later. The material reports a 96.2% score on the GraphWalks long-context benchmark, versus 84.2% for Gemini 4 Argon. Introductory prices are $2 per million input tokens and $10 per million output tokens, rising later to $4 and $20.
6 more sources reported this
- Latent Space AI
- 量子位 AI
- ginobefun (@hongming731)
- IT之家 科技新闻
- The Verge AI
- MarkTechPost AI 研究报道
Google opens Connected Apps platform preview for apps in Gemini and Search
Google has opened a public preview of its Connected Apps platform to U.S. developers who join the waitlist. Developers can submit apps and choose Gemini, Google Search, or both. Gemini already has more than 50 connected apps, which are set to reach Search over the next few weeks. Meditation app Calm is also due to launch in both.
Oct 28
NVIDIA Adds RTX Spark Laptop Features and Releases Nemotron Speech Model
RTX Spark laptops are available from six manufacturers, starting at $2,599. NVIDIA has released Nemotron 3.5 Transcribe for transcription, speaker labeling and translation, and added GPU resource controls in NVIDIA App. Liquid AI has introduced open weight LFM3 models; S-GAME’s Phantom Blade Zero is scheduled to launch at the stated time; and RTX Remix 1.6 supports Minecraft Java Edition 1.21.1.
7 related updates
- Jensen Huang · Statement · 黄仁勋在 GTC 柏林主题演讲中介绍英伟达的 AI 工厂、智能体、开放模型和物理 AI 技术，并表示欧洲正在建设 AI 经济。
- NVIDIA · Release · 英伟达推出开放权重语音模型 Nemotron 3.5 Transcribe，并已在 Hugging Face 提供该模型。
- NVIDIA · Release英伟达在 NVIDIA App 中推出资源管理功能，供用户分配前台应用与后台智能体使用的 GPU 资源。
- NVIDIA · Release英伟达发布 RTX Remix 1.6，新增对 Minecraft Java Edition 1.21.1 的支持。
- NVIDIA · Change英伟达的 RTX Spark 笔记本现已由 ASUS、Dell、HP、Lenovo、Microsoft 和 MSI 提供销售。
- NVIDIA · Release英伟达推出 Nemotron 3.5 Bolt E4B 4GB 模型，供本地语音助手、游戏角色和嵌入式界面使用。
- NVIDIA · Release英伟达的 DGX Spark 64GB 现已通过制造商合作伙伴开放购买。
NVIDIA releases Nemotron 3.5 Transcribe and the Granary v2 multilingual dataset
NVIDIA’s open-weight Nemotron 3.5 Transcribe is available on Hugging Face and ModelScope. It transcribes 40 languages and combines speaker attribution, translation and audio question answering in one model. The accompanying Granary v2 dataset covers 463,062 hours across 48 languages. In a medical speech adaptation case study, fine-tuning reduced word error rate on held-out speakers by 45.8% relative to the best prompt. The model is built for recorded audio, not live streaming.
2 more sources reported this · 3 related updates
- MarkTechPost AI 研究报道
- TechCrunch AI
Five vendors begin selling RTX Spark laptops as NVIDIA updates speech model and GPU controls
RTX Spark laptops from ASUS, Dell, Lenovo, Microsoft and MSI are now on sale, starting at RMB 18,999. NVIDIA has released the open-weight Nemotron 3.5 Transcribe model, which handles transcription, speaker labeling and translation in one model and supports transcription in 40 languages. NVIDIA App now lets users allocate GPU resources between foreground apps and background agents. Liquid AI's LFM3 series has also gained support for local use on RTX Spark.
5 related updates
- NVIDIA · Release英伟达发布开放权重语音模型 Nemotron 3.5 Transcribe，可用于转录、说话人标注和翻译。
- NVIDIA · Change英伟达为 NVIDIA App 新增资源管理功能，供用户在前台应用和后台智能体之间分配 GPU 资源。
- NVIDIA · Release英伟达发布 RTX Remix 1.6 版本，新增《我的世界》Java 版渲染支持及 RTX Spark 原生支持等功能。
- NVIDIA · Release英伟达推出 Nemotron 3.5 Bolt E4B 4GB 模型，其 4 位量化版本可支持自然对话和多步骤智能体任务。
- NVIDIA · Release英伟达通过 OEM 合作伙伴开售 DGX Spark 64GB 版。
Amazon unveils five Alexa+ home devices, opens Echo Show 16 preorders
The new devices span speakers, an alarm clock, a photo frame and a display, all with the latest Alexa+. Echo Studio with Home Camera lets the assistant identify objects in front of its camera, while Echo Time can wake users during a detected lighter sleep stage 10 to 60 minutes before their set alarm. Echo Show 16 starts at $499.99 and is available to preorder, with shipping expected later this year.
Volkswagen Group China unveils locally developed VEGA technology portfolio
VEGA covers vehicle platforms, an electrical and electronic architecture, and advanced driver assistance. Its CMP platform supports battery electric and plug-in hybrid vehicles. Volkswagen China’s Hyper Sense 8 driver assistance system supports urban and highway navigation assistance and parking assistance, with seven new electrified models planned to receive it in 2026.
1 more source reported this
- IT之家 科技新闻
HP, ASUS and Acer reportedly move some PC production back to China
Nikkei Asia reports that HP, ASUS and Acer are moving some PC production from Thailand and Vietnam back to China and working with new contract manufacturers as AI demand drives up the cost of memory and other components. Dell is evaluating a similar move. Omdia data shows that the bill of materials for a category of mainstream PCs rose from $435 in the first quarter of 2025 to $699 in the third quarter of 2026.
Microsoft launches MAI-Image-2.7-Preview for image editing
Microsoft has launched the MAI-Image-2.7-Preview image editing model, which users can try on Arena, Artificial Analysis and MAI Playground. It currently edits existing images using text instructions; text-to-image generation is not yet available. Microsoft says it improved over MAI-Image-2.6 across all 10 use cases and seven editing categories in an Artificial Analysis evaluation.