All AI updates
TechCrunch · AIFeatured7474
TechCrunch · AIOura acquires AI healthcare startup Counsel Health for $625 million
Oura announces acquisition of AI healthcare startup Counsel Health for $625 million, which provides AI-based medical consultations and doctor services, with its functions integrated into the Oura app. The transaction aims to expand services to preventive, emergency, and specialized chronic care, while Oura's revenue increased by 153% year-on-year to $533 million, reaching $1.75 billion in revenue for the fiscal year 2026. Counsel's CEO will join Oura as the Chief Clinical AI Officer.
The DecoderNvidia's GR00T 2 model learns tasks by observing human actions through video and imagines itself as a robot
Nvidia's GR00T 2 AI model learns new tasks for robots by analyzing human video actions without remote control, trained on 400,000 hours of video data, with lab test success rates of 10-80%, and combined with GPT-6 Astra achieves 97% success rate for toy model assembly, but tactile processing and language model latency remain challenges.
Databricks BlogDatabricks integrated Unity Catalog Metric Views into Power BI: a unified governance semantic layer solution.
Databricks released Unity Catalog Metric Views integration with Power BI functionality, extending the unified governance semantic layer to Power BI, allowing each Power BI report to use managed UC metric views the same as AI/BI dashboards and Genie, business users can build reports and explore metrics autonomously, data teams only need to maintain a single definition in Databricks.
Google AI BlogGemini Intelligence new features launch on more devices
Gemini Intelligence is now available on more devices, including the Motorola Signature 27 and Pixel 10. New features will expand capabilities on these devices, helping users complete more tasks. Create My Widget is launching soon, allowing users to create personalized widgets with prompts; task automation will expand to the UK, India, and Japan.
AI as Normal TechnologySTABL research organization is positioned in the key quadrant of AI impact.
The non-profit research organization STABL believes that advanced AI has transformative potential, but the speed of social transformation will be limited by human and organizational change, not technological progress. The organization advocates for focusing on the downstream bottlenecks of AI's impact, located in a two-dimensional quadrant that emphasizes both technological potential and impact bottlenecks, aiming to fill the current gap in thinking and promote research on institutional change and social resilience.
OpenAI News5050ChatGPT launches new features for enterprise
ChatGPT new features help businesses manage information, provide commercial display insights, and convert interest into action.
Latent SpaceNaiveAI's Naive-N0.5-Flash model generates 2,100 tokens per second, how is its programming capability?
Naive-N0.5-Flash is an open-source coding model developed by NaiveAI, achieving a high throughput of 2,100 tokens per second by improving Xiaomi's MiMo-V2.5 and integrating the NaiveRT engine.
The Verge · AISame story · 6 outlets6565Anthropic has released the AI office suite Claude Meetings, Sheets, and Whiteboards.
Anthropic has launched an AI office suite for non-technical knowledge workers, including Claude Meetings, Claude Sheets, and Claude Whiteboards, supporting features like meeting notes, shared whiteboards, spreadsheet editing, and presentation video generation, all currently in Beta testing. Some tools are available to paid plans, and Claude Data has been officially opened to all paid users.
Same story · 6 outlets- Claude · YouTube · Claude is launching whiteboard, spreadsheet, and data analysis features.First
- TechCrunch · AI · Anthropic is launching Claude Meetings to transform work discussions into plans and tasks.
- The Decoder · Anthropic has added collaboration features to Claude, launching plugins, spreadsheets, whiteboards, and an upcoming meeting notes feature.
How the accounts differ:The Verge states Claude Data is open to all paid users, while TechCrunch and The Decoder do not mention this.
The Verge · AI6565The independent game community is controversial about the use of AI.
Some judges at the Independent Games Festival (IGF) refused to vote for games using generative AI, while the developer of 'Big Walk' stated they used AI 'cautiously' in the later stages of game development to solve technical challenges, sparking intense community discussions about the impact of AI in game development, with no consensus yet.
The Verge · AIChatGPT launches verified AI 'Business Agents'
ChatGPT will allow users to chat with 'business agents' representing brands to help with comparative shopping. This feature is first rolled out to 'eligible' e-commerce and online service businesses in the US as part of the enterprise feature suite announced by ChatGPT today, entering Beta testing. Businesses can declare their organization and complete verification through the Business Hub feature to ensure the authenticity of the agents, while chats with business agents will be clearly marked and separated from the main ChatGPT conversation.
TechCrunch · AISame story · 3 outlets6565Ring introduces home security drone and commercial security robot
Ring introduces home security drone Ring Drone Cam and commercial security robot Ring Rover Cam, priced at 270 dollars and 1000 dollars respectively. The drone can autonomously patrol indoor environments and investigate when an alarm is triggered, equipped with 2K video and night vision features; the commercial robot can patrol along set routes, supports 4K video streams and fisheye camera, achieving autonomous navigation and map building through the AI Pilot system.
Same story · 3 outlets
The DecoderSame story · 12 outletsGoogle releases the Gemini 4 Carbon model
Google has released the Gemini 4 Carbon model, showing better reasoning performance than Fable 5.2 in the Artificial Analysis test, complementing rather than replacing Gemini 4 Argon, which is more suitable for most software engineering tasks.
Same story · 12 outlets- Artificial Analysis Articles · Gemini 4 Carbon topped the Artificial Analysis Intelligence Index v5First
- Google DeepMind · Google DeepMind releases the Gemini 4 Carbon cutting-edge model
- MarkTechPost · Google DeepMind released Gemini 4 Carbon, emphasizing network security defense and coding.
- Ars Technica · AI · Google has released Gemini 4 Carbon: a cutting-edge AI model that cannot be used immediately.
AWS Machine Learning Blog
Tomasz TunguzKC, the Chief Human Resources Officer at OpenAI, discussed the impact of AI on organizational change during Office Hours.
OpenAI's Chief Human Resources Officer KC states that terms like 'junior' and 'senior' are now meaningless because experience becomes outdated quickly. He believes AI models are like brains, needing to be combined with exoskeleton tools like Codex or Claude Code to achieve skills and cross-agent context management. When everyone can leverage AI to overcome their limitations, humans still rely on emotional intelligence and integrity as core competencies.
Where's Your Ed AtFeatured8080Ed Zitron's long article questions NVIDIA's GPU shipments to China via Megaspeed
Ed Zitron writes about the relationship between NVIDIA CEO Huang Renxun and Megaspeed CEO Alice Huang, as well as Megaspeed's procurement of approximately $4.8 billion NVIDIA GPUs via its Malaysian subsidiary Speedmatrix.
Why it matters: The article links the controversy over NVIDIA's exports to China with the personal lives of executives and investigations into lobbying interference, allowing readers to understand the context of its revenue disclosure controversy.
Codex releases00GitHub Copilot 0.167.1 released
GitHub Copilot 0.167.1 is released, providing features such as code review, vulnerability detection, and secret scanning.
Google Developers BlogGoogle AI Edge Gallery introduces on-device speech recognition and text-to-speech capabilities
Google AI Edge Gallery has launched on-device automatic speech recognition (ASR) and text-to-speech (TTS) capabilities, enabling fully offline operation of multilingual ASR and natural TTS models on phones and wearable devices.
Hugging Face BlogHugging Face Hub supports payment using Google Cloud accounts
Users can now associate the Hugging Face organization with their Google Cloud account via Google Cloud Marketplace, allowing Hugging Face usage to be billed to the Google Cloud account instead of credit card payment.
Claude · YouTubeClaude enhanced interactive connector function, can use the full tool version directly within the application
Claude adds interactive connector functionality, allowing users to directly open the full version of the tool within applications like Adobe, Atlassian, and Box for collaboration. The video demonstrates adding a plan summary in Confluence, applying a dither effect in Adobe, and checking flat design standards in Box, among other operations, and supports users to modify it themselves or have Claude assist with it. This feature has started rolling out on web and desktop versions.
Claude · YouTubeSame story · 6 outletsClaude is launching whiteboard, spreadsheet, and data analysis features.
Claude Whiteboards and Claude Sheets enter beta, Claude Dashboards officially launch as Claude Data
Same story · 6 outlets- The Verge · AI · Anthropic has released the AI office suite Claude Meetings, Sheets, and Whiteboards.
- TechCrunch · AI · Anthropic is launching Claude Meetings to transform work discussions into plans and tasks.
- The Decoder · Anthropic has added collaboration features to Claude, launching plugins, spreadsheets, whiteboards, and an upcoming meeting notes feature.
How the accounts differ:The Verge states Claude Data is open to all paid users, while TechCrunch and The Decoder do not mention this.
TechCrunch · AISame story · 6 outletsAnthropic is launching Claude Meetings to transform work discussions into plans and tasks.
Anthropic has launched Claude Meetings, integrating Claude into video conferencing to generate plans, create timelines, answer questions, and access connected apps in real-time during discussions; simultaneously, Claude Whiteboards and Claude Sheets were released, supporting team real-time collaboration for visual planning and co-editing spreadsheets, both in Beta testing.
Same story · 6 outlets- Claude · YouTube · Claude is launching whiteboard, spreadsheet, and data analysis features.First
- The Verge · AI · Anthropic has released the AI office suite Claude Meetings, Sheets, and Whiteboards.
- The Decoder · Anthropic has added collaboration features to Claude, launching plugins, spreadsheets, whiteboards, and an upcoming meeting notes feature.
How the accounts differ:The Verge states Claude Data is open to all paid users, while TechCrunch and The Decoder do not mention this.
Meta EngineeringMeta open-sources Triton Low-Level Extensions (TLX): a hardware-native GPU compiler optimized for large-scale production ML
Meta open-sources Triton Low-Level Extensions (TLX) compiler, enhancing GPU performance by introducing explicit warp-group and cluster-level orchestration, supports RecSys training/inference and LLM pre-training, achieving end-to-end performance improvements over standalone libraries on NVIDIA and AMD GPUs.
Meta EngineeringMeta launches Your Algorithm tool to make Facebook recommendation content more transparently personalized
Meta has launched the Your Algorithm tool, allowing users to edit interest lists ("Interested In"/"Not Interested In") in natural language to adjust Feed and Reels recommendations in real-time. The system uses LLM to process user preferences and integrate user behavior, achieving responsive adjustments through retrieval, reordering, and pushing optimization stages. New interest content will appear quickly but will not dominate recommendations.
LangChain BlogWorkday how to build an AI intelligent agent platform with LangGraph and LangSmith
Workday has utilized LangGraph for agent orchestration and LangSmith for tracking and evaluation, reducing prototype development time from 5 months to within 1 day with a 4-person team by 2025 and supporting production environment deployment. Its Python SDK 'Workday LangChain' encapsulates LangChain core components and routes through internal services (including LLM gateway, RAG retrieval) while meeting enterprise-level security and compliance requirements.
PyTorch BlogExecuTorch WebGPU backend optimization Muse Spark Open performance on TPU v7
ExecuTorch WebGPU backend allows PyTorch models to run in WebGPU browsers without converting framework formats, supporting multiple models such as Llama 3.2, Qwen3, and Gemma 3.
The DecoderSame story · 4 outletsSame story · 4 outlets- Google DeepMind · Google DeepMind releases Gemini Robotics 2.5 vision-language-action modelFirst
- Ars Technica · AI · DeepMind releases Gemini Robotics 2.5 and open-source security benchmark "ASIMOV-jailbreak"
- ITHome · AI · Google's strongest VLA robot model, Gemini Robotics 2.5, debuts, can tie bows, already on the Mercedes production line
AWS Machine Learning BlogAmazon Bedrock Observes Data Lake: Building a Centralized Call Log Analysis Platform
Amazon launched Bedrock Observational Data Lake, which transmits original model invocation logs from linked accounts to a central account via CloudWatch. Processed by Data Firehose and converted to Parquet format for storage in S3, it ultimately provides a unified query interface through Athena.
AWS Machine Learning BlogHow TCS uses Amazon Bedrock AgentCore to build an agentic test orchestrator
TCS has built an Agentic test orchestrator for BFSI clients, transforming requirements into executable controlled tests through Amazon Bedrock AgentCore, breaking down tests into dedicated tasks and executing them using their respective agents, tools, and validation rules, while meeting project-specific context, API, and browser UI testing, tenant-level security isolation, and repeatable test path requirements, significantly improving test throughput and allowing engineers to focus on risk analysis and exploratory testing.
Sierra Blog3535How did Sierra designers use AI to complete the Mossy project in two weeks?
Sierra designers used AI to complete the Mossy project in two weeks, a project aimed at designing a visible voice assistant.
MiniMax NewsFeaturedSame story · 3 outlets8080MiniMax releases open-source image generation model H3-Image
MiniMax has released an open-source weight image generation model based on H3, H3-Image, which supports text-to-image, reference image generation, and image editing, capable of generating 4096 × 4096 4K images within 10 seconds, and supports up to 16 reference images.
Same story · 3 outletsWhy it matters: The official provides 4K generation speed, reference image quantity, and pricing range, which can be used to judge the gap between open-source image models and closed-source solutions.
OpenRouter BlogHow to get structured output from any LLM: JSON Schema, tool calls, or decision models
OpenRouter tested three methods for obtaining structured outputs from LLMs, validated on nine models supporting structured outputs and only tool-supported GLM 5.3 FlashX, where 209/216 responses matched strict JSON Schema.
OpenRouter BlogHow to build a voice AI agent (based on OpenRouter API and Python)
Open-source multimodal model Claude Code v2.1.159 released, supporting multilingual and code generation.
OpenRouter BlogOpenRouter Test: Claude Opus 5.5 is the best OCR model for scanning documents
Claude Opus 5.5 identified 262 out of 264 key values in OpenRouter testing, leading with a price of $16.42 per thousand pages, Mistral OCR ranked second with 258 key values and a cost of $2.02 per thousand pages, and Ling 3.0 Flash VL performed best with 254 key values and a cost of $0.04 per thousand pages due to disabled reasoning.
n8n BlogSame story · 12 outlets6262Google Gemini 4 Carbon is released in n8n, enhancing the security, reliability, and efficiency of agents and workflows
Google Gemini 4 Carbon has officially launched on n8n, being Google's strongest model, scoring 53.9% on AutomationBench. It significantly improves in security (SEC-bench Pro 88.2%, ExploitGym 54.8%, CWE-bench 71.0%), long task processing (internal testing lasting several days), and reasoning, and has zero vulnerability to Gray Swan prompt injection attacks (0.0%).
Same story · 12 outlets- Artificial Analysis Articles · Gemini 4 Carbon topped the Artificial Analysis Intelligence Index v5First
- Google DeepMind · Google DeepMind releases the Gemini 4 Carbon cutting-edge model
- MarkTechPost · Google DeepMind released Gemini 4 Carbon, emphasizing network security defense and coding.
- Ars Technica · AI · Google has released Gemini 4 Carbon: a cutting-edge AI model that cannot be used immediately.
Codex releases00GitHub Copilot 0.169.0-alpha.9 released
GitHub Copilot released version 0.169.0-alpha.9, without mentioning specific update content.
NVIDIA NewsroomNVIDIA expands open-source commitment: donates OpenShell and launches new AI infrastructure
NVIDIA has contributed OpenShell, an open-source autonomous agent secure runtime, to the Cloud Native Computing Foundation (CNCF) and launched new tools such as NVIDIA Kubernetes Engine and NVIDIA Kubernetes Recipes to help teams deploy and manage GPU infrastructure on Kubernetes.
Factory NewsSecure Factory: Connecting Security Discovery, Remediation, and Validation into a Continuous Workflow
Security Factory connects the signal, classification, and remediation three stages to help engineering teams transform security discoveries into test fixes. It integrates existing tools and contextual information, automatically coordinates investigation tasks, generates a risk-prioritized remediation list based on validation, and simplifies the entire process from discovery to remediation.
Comfy BlogMiniMax H3 Image supports ComfyUI
MiniMax H3 Image now natively supports ComfyUI, offering 4096×4096 high-resolution output, supports text-to-image generation and reference image editing, and can also be run directly through Partner Nodes.

























