This issue collects 29 AI updates from 2026-11-09, led by Cloudflare and Tools for Humanity Propose Identity Framework for AI Agents, Nvidia board opens independent review of Jensen Huang's ties to alleged Chinese spy, Report Says Anthropic Is Building AI Systems Twice as Fast as Last Year, GammaDX index finds only 12 of 50 Australian websites are agent ready.
Cloudflare and Tools for Humanity have proposed Trusted Agent Delegation, an identity framework that lets websites verify who is behind an AI agent without blocking agents wholesale.The plan combines four tiers of user identification with reputation and accountability mechanisms, and the partners aim to bring it to the IETF as bot traffic surpasses half of the internet.
Australian digital experience agency GammaDX has launched an AI agent experience design practice alongside an index of how ready local websites are for AI agents.Of 50 sites tested across five sectors, only 12 were rated agent ready, with trust and governance the weakest category at 56%.
A community project called dsh keil mcp exposes a local Keil MDK installation, UV4 compiler and hardware debug probes to coding agents through MCP, letting them discover projects, build, edit source, flash firmware and read registers.The MIT licensed Alpha 0.1.
LG AI Research said on November 9 that it is opening the API for K EXAONE 2.5 free of charge, announcing the move at NVIDIA AI Day Seoul 2026.The model is a technology validation build created by further training the second generation model released in August, and LG says combined performance across 24 metrics rose more than 70%.
QbitAI's November 9 hands on report describes how LobeHub lets supervisor agents dispatch work, coding agents write code and review bots catch bugs, with the collaboration platform now open sourced on GitHub at roughly 84,400 stars.The team claims a three to five times efficiency gain, while humans keep final sign off.
NVIDIA and South Korea's Ministry of Science and ICT are widening their sovereign AI collaboration to cover AI semiconductors, data center infrastructure, talent, startups, models, agents and security.The partnership also opens a path for Korean NPU and XPU companies to explore joining NVIDIA's NVLink Fusion ecosystem.
Wirken is an open source gateway that sits between messaging apps and AI agents and assumes any agent can be turned against its owner.It isolates each channel in its own process, keeps credentials in an encrypted vault, runs shell commands in a networkless sandbox and records every action in a signed, hash chained log.
TokenRhythm has opened a limited beta for NeoHorse 1.5 Flash Preview, a 180B parameter model positioned for agent workloads, according to Pandaily.The model is being offered in preview form, with details on access scope and availability still thin.
DeepSeek's AI agent push is gaining pace, with its Harness product now exceeding 1.5 million daily active users, according to the South China Morning Post.The figure shifts the conversation from model benchmarks to real usage at the consumer facing end of the agent race.
Qwen Office launched version 2.0 on November 9, led by a Collaboration feature the company calls the industry's first to support multiple people and agents in one shared workspace.The release also adds human AI co creation, a material space, automated tasks and three device sync, all now generally available.
A first person account published by Business Insider describes how Robert Farlow, founder of the San Francisco AI startup Plato, handed office move coordination to OpenAI's Dot agent and watched it send roughly 300 emails across movers, insurers and building managers in a few days.Farlow shut the agent down and now reviews every AI drafted message before it goes out.
QbitAI has published a hands on test of Tencent Hunyuan's new open source model Hy4, placing it in the top tier of open models.The test paired Hy4 with WorkBuddy to judge multimodal agent delivery on real tasks rather than single turn benchmark scores.
Tarahum Charity Foundation launched Rahma, an AI agent on its website, on November 9 to guide visitors to the information and services they need and answer common questions about giving.The foundation says the agent absorbs routine inquiries so staff can focus on cases that need personal follow up, particularly People of Determination and senior citizens.
Nvidia's board has launched an investigation led by independent directors into CEO Jensen Huang's relationship with Alice Huang, a Megaspeed executive whom a memo circulating among U.S.policymakers alleged was recruited by China's Ministry of State Security.
NVIDIA's blog argues that as AI agents take on more work, capable open models and security have to advance together.Hyundai Motor Group is exploring NVIDIA NemoClaw blueprints and the OpenShell runtime, while AIM Intelligence is integrating its Starfort platform with OpenShell.
NVIDIA and South Korea's science ministry hosted an AI startup forum at COEX in Seoul on November 9, meeting more than 20 Korean companies across foundation models, physical AI, AI infrastructure and AI chips.NVIDIA announced an investment in interconnect startup Point2 and said more investments are planned in Motif Technologies, RLWRLD and Upstage, without disclosing amounts or a timeline.
Palo Alto Networks is in talks to acquire AI agent security startup Clutch Security for about $500 million, according to AI Insider.The report describes ongoing negotiations rather than a signed deal, and neither company has confirmed the terms.
Zoho has upgraded Zoho Campaigns, combining email, SMS and WhatsApp campaigns in one workspace and adding native WhatsApp marketing, frequency capping and Model Context Protocol support so compatible AI agents can coordinate campaign preparation, segmentation and analysis.The platform is available globally at unchanged pricing, with paid plans starting at 2,300 Nigerian naira per month.
Space and Time announced Space and Time Enterprise on November 9, letting companies run AI tools over private databases and prove the results are correct without handing over the underlying records.The platform extends its Proof of SQL work using zero knowledge proofs and needs no blockchain, GPUs or specialized hardware, running on ordinary laptops and servers.
Disney and Adobe have signed a multi year agreement to develop private AI models, according to a FourWeekMBA report.The deal points to large content companies moving from off the shelf generative tools toward models built and controlled for their own operations.
Anthropic is developing its AI systems about twice as fast as a year ago, according to an Associated Press report dated November 8, even as chief executive Dario Amodei has called for a slower pace on model capabilities.The same reporting points to disclosures in the company's IPO filing, including a pause of more than a month in training its next flagship model and previously undisclosed safety evaluation failures.
NVIDIA's blog outlines how South Korea's physical AI ecosystem is advancing, with local industry leaders applying accelerated computing, AI and simulation across semiconductor manufacturing, robotics and autonomous vehicles.Samsung's semiconductor engineering work is placed at the front of the story.
ByteDance has released Seedream 5.5, a new image creation model whose two headline features are up to 100 successive edits on a single image and direct transparent background output.Both target the practical loops of iterative retouching and asset post processing rather than one shot image quality.
Google's AI Plus subscription will stop offering access to Gemini Pro models on November 13, according to a report carried by Ammon News, affecting the company's entry level paid AI plan.The report does not explain the reasoning or say what subscribers get instead, leaving only a few days before the change takes effect.
A short NVIDIA blog item says Dynamo Recipes can bring South Korea's leading open models into production.The point is the deployment path rather than a new model, and the source gives no model list or performance figures.
At CoRL in Austin, NVIDIA released GR00T 2, an open world action model built on Cosmos that completed 40 of 50 evaluated tasks without task specific demonstrations.The accompanying Isaac Lab 3.0, Newton physics integration and data tools push simulation based robot training toward industrial deployment.
NVIDIA, Korean chip designer Rebellions and telecom operator SK Telecom have announced a collaboration.The available summary stops at the headline, with no stated scope, technical detail, investment size or timeline.
NVIDIA is reportedly looking to use Brightspeed's telecom central offices as locations for its inference network, according to Data Center Dynamics.If confirmed, it would mark a shift toward distributing AI compute into existing telecom facilities closer to users instead of concentrating everything in large data centers.
Apple has cleared a key regulatory step in China for bringing Apple Intelligence features to iPads and Macs, according to a TipRanks report, sending the company's shares slightly higher.The report does not name the approving authority, list the cleared features or give a launch timeline.