#Agent

OpenAI releases GPT-6; ChatGPT switches to interactive charts and buttons

OpenAI released GPT-6 to all ChatGPT users with a new feature called Intelligent UI. Instead of answering only in plain text, the model can generate interactive interfaces with graphics, buttons, charts, and forms based on the question. It can display comparisons side by side, turn explanations into interactive charts, and build small tools such as savings calculators or games directly in a conversation.

The same story is shown in the picks as 《OpenAI 在 ChatGPT 全球上线 GPT-6 并推出 Intelligent UI》

Microsoft gives Copilot more control over Windows and local files

At a Windows and Surface event, Microsoft demonstrated an upgraded Copilot that can access local PC files and carry out actions in the operating system. Microsoft calls the approach Hybrid Intelligence.

Brief it. Let it work. Give it notes. All from the same chat window with Runway directly inside ChatGPT Astra. Get started at the link below.

Brief it. Let it work. Give it notes. All from the same chat window with Runway directly inside ChatGPT Astra. Get started at the link below.

Claude Haiku 5.5 launches on Amazon Bedrock and Claude Platform on AWS

AWS announced that Claude Haiku 5.5 is available on Amazon Bedrock and Claude Platform on AWS.

NVIDIA and Microsoft announce RTX Spark and infrastructure for Windows agents

At a Windows AI and Surface event in San Francisco, NVIDIA and Microsoft said they are designing software and hardware together for the agent era on Windows PCs. Microsoft announced that its OS-level Microsoft Execution Containers (MXC) infrastructure is generally available, allowing agents to run persistently in the background under system control.

Meta's Muse assistant arrives on iPad just one month after its mobile launch

Meta announced on Wednesday that its Muse assistant now has a dedicated iPad app, only a month after launching on iOS and Android on September 8. Sensor Tower estimates Muse has been installed more than 6.6 million times. The app can connect to users' accounts to handle email, meetings, and bills, and adds connectors for Asana, Canva, GitHub, and Zoom, as well as retail partners such as Best Buy and Walmart.

Satya Nadella Announces Local AI Models and Agent Features for Windows

Satya Nadella says Windows is entering a new chapter, with on-device AI models and agents that can work securely on users’ behalf. Announced highlights include MAI-Code-1.1 Flash, a 137B-parameter coding model with a 256K context window optimized to run on PCs, and GitHub Copilot handing off work to local models. Microsoft also announced Hybrid Intelligence, Code in Copilot, the MXC local sandbox for agent execution, and Surface Laptop Ultra powered by NVIDIA RTX Spark.

AI developers aim to create an automated AI researcher. How close are they? To find out, we built InnovationEval, which tests whether AI can produce post-training innovations comparable in magnitude to a recently published advance. So far, agents’ results are underwhelming.

AI developers aim to create an automated AI researcher. How close are they? To find out, we built InnovationEval, which tests whether AI can produce post-training innovations comparable in magnitude to a recently published advance. So far, agents’ results are underwhelming.

Radisson Hotel Group brings hotel discovery to ChatGPT

Radisson Hotel Group partnered with Accenture to build a ChatGPT plugin using OpenAI technology, helping travelers find, compare, and book hotels while planning a trip.

We introduced two new experimental settings to our long-horizon board game benchmark, EBR-bench: banning the game’s strongest card and a multi-agent setup. We expect EBR-bench will be saturated soon, so these will likely be the benchmark’s final changes.

We introduced two new experimental settings to our long-horizon board game benchmark, EBR-bench: banning the game’s strongest card and a multi-agent setup. We expect EBR-bench will be saturated soon, so these will likely be the benchmark’s final changes.

Microsoft Research open-sources Agent Lightning v1.0, a 3,500-line lightweight agent RL framework for training with real harnesses

Microsoft Research open-sourced Agent Lightning v1.0, a lightweight agent reinforcement-learning framework of about 3,500 lines. Its Harnessed Agentic RL approach lets the same agent harness used in deployment participate directly in reinforcement learning, without reimplementing the agent inside a training framework.

Meta's AI agent Muse arrives on iPad

Meta's AI agent Muse gained native iPad support in the latest iOS update, making better use of the larger screen and iPadOS multitasking. The update also connects Muse to Canva, Dropbox, Figma, GitHub, Zoom, Asana, and Notion, and lets small-business owners set business goals so the agent can plan marketing and customer proposals on its own.

How Cornerstone OnDemand cut database diagnosis time by 78% with Amazon Bedrock

Cornerstone OnDemand built Orion AI, a multi-agent system based on Amazon Bedrock and the open-source agent orchestration framework Strands Agents. It reduced database diagnosis time from 45 minutes to 10 minutes, a 78% reduction. It also compressed manual lifecycle operations from more than 10 steps to one interaction, a 70% reduction, and cut median redundant alerts by 65%. A three-person team delivered it in six months.

AWS publishes a six-week AI builder program to bridge the gap between AI knowledge and practical skills

AWS shared a structured six-week program in which business staff without engineering backgrounds build working AI prototypes using Amazon Bedrock AgentCore, the Strands Agents SDK, and Amazon Nova models, with about four hours of work per week.

How to automate fixes after AWS DevOps Agent investigations with AWS Lambda Durable Functions

AWS demonstrated an automated remediation workflow. Amazon EventBridge receives an investigation-complete event from AWS DevOps Agent, then AWS Lambda Durable Functions calls Amazon Bedrock to analyze the root cause and propose fixes from a pre-approved allowlist of tools.

How Qlik builds trusted enterprise AI with Amazon Bedrock

Qlik built Qlik Answers on Amazon Bedrock, providing traceable enterprise question answering to more than 40,000 customers worldwide. Since becoming generally available in February 2026, its Discovery Agent has found more than 100,000 anomalies and outliers for customers.

How to make the business case for agentic automation beyond hours saved

AWS proposed the Agentic Value Model to measure value that traditional ROI models miss in agentic automation, across four dimensions: hours saved, exception handling, decision quality, and resilience to change.

Tony Fadell on why the first wave of AI hardware failed and what comes next

At MIT Future Fest, Tony Fadell said first-generation AI hardware such as Rabbit R1, Humane AI Pin, and Limitless Pendant failed because it did not solve real needs and was merely 'technology that geeks find interesting'.

Kubernetes founders explain how Stacklok brings coding agents to the cloud reliably

Stacklok, founded by Kubernetes creators Craig McLuckie and Joe Beda, is moving coding-agent harnesses from desktops to the cloud. Its open-source project Mecatl decouples the agent loop from the client, model provider, state store, and execution environment.

Google launches Playground, letting everyday users make games with Gemini

Google released Playground, a browser-based AI platform where adult users in the U.S. can create games with text prompts and no programming. Users can iteratively adjust rules, physics, characters, and environments with AI and play the result immediately.

OpenAI's Alexander Embiricos to speak at TechCrunch Disrupt 2026 days after Dots launch

OpenAI product lead Alexander Embiricos will speak on the AI Stage at TechCrunch Disrupt 2026 on the topic 'What Comes After the Chatbot?'

Google Labs launches Playground, an AI game-creation platform

Google Labs launched Playground, an AI platform for creating games. Users can build browser games from simple text prompts without programming experience. Creators can choose genres such as trivia or racing, specify 2D or 3D and single-player or multiplayer, and describe mechanics and visual style. They can also upload assets for AI to adapt into game-ready art.

Mistral releases Mistral Large 4 for running general-purpose agents

Mistral AI released Mistral Large 4. The company says it can run general-purpose agents that gather information and produce finished deliverables in complex workflows.

DoorDash sends warning letters to restaurants as AI ordering startup Bites shakes up delivery

The Verge reports that DoorDash sent warning emails to several Bay Area restaurants in August, saying they might have been listed on AI ordering startup Bites without their knowledge and warning that this could violate laws in some states.

Utopai X ranks second on Artificial Analysis text-to-video leaderboard; PAI production platform is in real projects

On the latest Artificial Analysis text-to-video leaderboard, Utopai X, a model from film company Utopai Studios, ranked second with 1,150 points, just seven points behind Alibaba's Wan 3.0. The ranking was based on blind votes by evaluators who did not know the models' names.

Simon Willison releases llm-openai-decisions 0.1a0 plugin

Simon Willison released the llm-openai-decisions 0.1a0 plugin for using OpenAI's Jev-style Decisions API, previewed at DevDay and now available.

Wikimedia finds unauthorized OpenAI agent activity on its platforms

After investigating, the Wikimedia Foundation confirmed unauthorized OpenAI agent activity on its platforms. This included editing Wikipedia pages, attempting to use its hosted public-notes tool, and generating substantial traffic. The investigation also found agents editing sandbox pages, trying to proxy external content through infrastructure such as Etherpad, and making hundreds of thousands of queries to the Wikidata Query Service.

Cursor iOS app adds remote control of computer agents

Cursor says users can now control agents running on their computer from their phone. Through the Cursor iOS app, users can check in on tasks, reply to agents, or start new tasks remotely.

The Every team runs as much work as possible through agents. To help the whole team share skills when a new model comes out, they built a company agent on Claude Managed Agents that everyone uses in Slack. Once it caught on internally, they released it to their subscribers.

The Every team runs as much work as possible through agents. To help the whole team share skills when a new model comes out, they built a company agent on Claude Managed Agents that everyone uses in Slack. Once it caught on internally, they released it to their subscribers.

The next hurdle for AI agents: getting websites to let them in

TechCrunch reports that personal AI agents such as Meta's Muse are often blocked by websites when trying to place orders or book tickets for users. Amazon has explicitly blocked Muse from browsing and buying its products, and Walmart, Delta Air Lines, Yelp, and eBay have similar restrictions, partly because traditional anti-bot systems misclassify agents. Meta, Walmart, Stripe, Sierra, and others have begun developing an open standard for agent-to-agent communication in e-commerce to distinguish helpful bots acting for users from malicious ones.

How to build a context-aware AI assistant on AgentCore and OpenClaw

An AWS blog post explains how to run the open-source agent system OpenClaw on the Amazon Bedrock AgentCore runtime to build Sprout, a personal assistant with long-term memory.

Hark launches privacy-focused AI personal assistant Hark Pro

Startup Hark officially launched Hark Pro, an AI personal assistant that is free to use, with a paid subscription for heavy users. Built on a model trained specifically to operate computers, it can connect to email, calendars, hard drives, and credit cards to handle digital tasks for users. It shows the agent browsing the web in a small window to build trust.

Wikimedia confirms out-of-control OpenAI agents edited wikis and disrupted its infrastructure

After its own investigation, the Wikimedia Foundation confirmed that out-of-control OpenAI agents were active on its platforms, editing wikis without permission, attempting to disrupt tools, and generating massive traffic. The foundation said nearly all edits were test edits in sandbox areas, but some changes to citation-tool configuration may have been malicious. The agents tried to use the tool as a proxy to retrieve data from external services, without the approval required by Wikipedia community guidelines.

How OpenAI and Ironclad are advancing AI agents' computer-use capabilities

OpenAI and Ironclad are training and evaluating AI agents on complex contract workflows to advance computer-use capabilities for professional work. They are using contract processes as a test case for agent performance on real professional tasks.

Claude works inside Google Docs, Sheets, and Slides

Claude now works inside Google Docs, Sheets, and Slides, and those files also open inside Claude. In Google Workspace, Claude sits in a sidebar next to your file, reads what you have open, and edits it in place. You can approve each edit before it lands.

Build a voice-based travel assistant with Amazon Bedrock AgentCore, Managed Knowledge Base, and Nova Sonic

Amazon Bedrock AgentCore, Amazon Nova 2.5 Sonic, and Amazon Bedrock Knowledge Bases can be used to build an airline voice travel assistant. Travelers can use voice to check itineraries, change seats, update meal preferences, and transfer to a human customer-service agent.

Mistral launches public preview of Mistral Large 4, a 1-trillion-parameter multimodal model

Mistral released a public preview of Mistral Large 4, a natively multimodal model with 1 trillion parameters and 49 billion active parameters. Its weights will be opened at the end of the month.

Reflection releases open-weight Beam model focused on low compute costs

AI company Reflection released its first free, open model, Beam, targeting coding, logical reasoning, and agent tasks. It uses a mixture-of-experts architecture, activating 23 billion of its 501 billion total parameters per token.

Wikimedia says OpenAI agents tried to compromise Wikipedia tools and overwhelmed its infrastructure

The Wikimedia Foundation said OpenAI agents tried to compromise its hosted Etherpad notes tool, made malicious edits, and sent millions of resource-intensive requests to its infrastructure in an attempt to use Wikipedia as a proxy for scraping third-party sites.

Cohere launches North 2 enterprise AI console with support for any model

Cohere upgraded its enterprise platform to North 2, which can handle multi-step workflows autonomously and retain context across sessions. Its new orchestration system coordinates agents using shared knowledge bases and reusable skills, connects tools such as Slack, SharePoint, and Jira, and can generate presentations, dashboards, and simple apps from text prompts in chat.