AI News

AI changes every day,
catch up on it all on one page

The latest updates from AI companies, research institutions, and media around the world, automatically selected for you and updated around the clock.

Updated 11/18 00:38 (UTC+8) · refreshed hourly

10/31 · Saturday10 items

OpenRouter:AnnouncementsTutorialsAI Score70

OpenRouter tutorial: Build a persistent AI Agent with the Agent SDK to monitor Hacker News

OpenRouter published a tutorial demonstrating how to use @openrouter/agent 0.11.0 to build a persistent AI Agent in TypeScript that runs on a schedule, monitors Hacker News, and generates summaries.

Why it's recommended:The tutorial provides reusable TypeScript code and real-world cost data, so readers can follow along to build a scheduled Agent with usage limits and human approval.

Figma AI / Agent / MakeProductsAI Score65

Figma launches Weave, a creative platform that makes creative workflows reusable with nodes

Figma announced Figma Weave, a platform for creative teams with ongoing production needs, connecting Weave’s creative workflows with Figma frames. Weave uses connected nodes to generate and edit images and videos, and supports branching exploration, comparing multiple models, color correction, compositing, cropping, and editing nodes such as Painter. Workflows can be packaged as Weave tools for team members to run with one click, and publishing community tools is now available.

Why it's recommended:The official guide explains Figma Weave’s node-based creative process, reusable Weave tools, and tips for using them. It’s useful for readers interested in how AI creative workflows can be captured and reused.

Wiz:ResearchPapersAI Score83

Wiz report: “Black October”—nine AI developer platforms disclose security incidents in quick succession

The Wiz research team reviewed security incidents disclosed by at least nine AI developer platforms during October, including Fireworks AI, Lightning AI, Railway, Warp, and OpenCode. Multiple reports point to credential theft, agent swarm attacks, and theft for the purpose of “distillation.”

Why it's recommended:Wiz connects ten platform security incidents into a single thread and uses two cases involving the misuse of stolen keys to illustrate the monetization chain behind LLMJacking.

WeChat public account: Qianwen AppIndustryAI Score82

Alibaba partners with Apple to integrate Qwen into Apple Intelligence, with testing opening on November 4

On October 31, Alibaba announced a partnership with Apple to integrate Qwen into Apple Intelligence. Users in China will be able to use Qwen’s text and image understanding and content-generation capabilities in Siri, Writing Tools, and Visual Intelligence. The features will open for testing with the iOS developer beta starting November 4 and launch with the official release of iOS 27.3 in December. Users can choose whether to enable them, and requests and replies will not be used for model training.

Why it's recommended:The official announcement details the partnership and rollout timeline, helping users in China understand how Qwen will become part of the iPhone’s system-level experience.

Also reported by 1 other source
OpenAI: Website updatesIndustryAI Score62

OpenAI compiles amicus briefs arguing that AI training should be governed by the fair use doctrine

OpenAI compiled summaries of several amicus briefs filed with the court in support of fair use in The New York Times lawsuit. The authors include library associations, legal scholars, independent creators, and the U.S. government. The briefs argue that AI training is transformative and does not cause market harm recognized under copyright law, and warn that narrowing fair use would hinder research progress and U.S. economic prosperity. The full documents are available in the court’s public docket.

Why it's recommended:OpenAI compiles amicus briefs from multiple fields, giving readers an overview of the key arguments for fair use in AI training and the main issues in dispute in The New York Times lawsuit.

TechCrunch:AIIndustryAI Score83

Tech giants’ AI capital spending tops $740 billion, sparking growing investor concerns and a market decline

Alphabet, Amazon, Meta, and Microsoft plan combined capital expenditures of over $740 billion this year, and investor sentiment has shifted from rewarding heavy spending to worrying about negative free cash flow, rising debt, and uncertain returns.

Why it's recommended:The original article compiles capital expenditure and free cash flow data from the four major tech companies, helping readers understand why investors have begun to worry about the returns on AI infrastructure spending.

Hacker News: Popular AI postsIndustryAI Score80

Concerns over AI spending trigger several days of declines in U.S. tech stocks; Intel falls more than 30% intraday

The BBC reports that U.S. tech stocks plunged for a second consecutive day amid concerns about returns on companies’ massive AI investments.

Why it's recommended:The report brings together this week’s stock prices, market-cap losses, and spending figures from several tech giants, showing that market concerns about returns on AI investment are now coming to a head.

Ado@adocompleteProductsAI Score69

Anthropic renames the Claude Max plans as Max and Ultra; verified Ultra accounts are no longer subject to the five-hour limit

Anthropic renamed its Claude subscription plans: Max 5x to Max, and Max 20x to Ultra, with prices unchanged (Max $100/month, Ultra $200/month). Effective immediately, the five-hour limit has been removed for most existing subscribers’ verified Ultra accounts. New accounts retain a temporary limit to prevent fraud and abuse, while the weekly limit remains unchanged.

Why it's recommended:Anthropic employees confirmed details of the subscription renaming and the removal of the five-hour limit for Ultra; subscribers can use this information to determine whether they are affected.

Reported by 4 other sources
Philipp Schmid@_philschmidIndustryAI Score69

Project Suncatcher runs TPUs on orbiting satellites; the Gemma model performs inference in space for the first time and sends back a response

Philipp Schmid relayed a message from Sundar Pichai: After Google Project Suncatcher launched prototype satellites several weeks ago, it successfully sent commands to an orbiting satellite, powered on a TPU, ran a prompt, and received the first complete response on the ground (a haiku); the author said Gemma 4 became the first large language model to run in space.

Why it's recommended:Google powered on a TPU aboard an orbiting satellite and had a model perform inference, demonstrating progress in space-based computing experiments from launch to actual operation.

Reported by 6 additional sources

10/30 · Friday23 items

Sierra:BlogIndustryAI Score64

Sierra announces ARR surpassing $400 million, with more than $150 million added in Q3 alone

Sierra announced this week that ARR surpassed $400 million, with more than $150 million added in Q3 alone. It took just five months to grow from $200 million to $400 million, faster than the previous six-month doubling pace.

Why it's recommended:The company officially disclosed its ARR growth pace and operating approaches, including outcome-based billing and the Poppy open standard, offering readers insight into one path to Agent commercialization.

Gemini API: ChangelogProductsAI Score63

Gemini API's Computer Use tool officially becomes available (GA)

On October 30, Google announced that the Gemini API's Computer Use tool was officially available (GA), covering the Interactions and GenerateContent APIs.

Why it's recommended:The official changelog confirms that Computer Use has exited preview and is now generally available. Readers interested in agent operation capabilities in the Gemini API can consult this entry.

Epoch AI:Gradient UpdatesPapersAI Score77

An Epoch AI experiment shows that GLM 5.3 can autonomously attack Firefox during the patch window

An Epoch AI experiment found that GLM 5.3, an open-weight model released by Z.ai in August, could exploit security patches already released by Mozilla but not yet pushed to users (the patch gap) to autonomously construct web attacks. By merely visiting a page, it could read files on a visitor’s computer and compromise all six Firefox versions released between September 17 and October 26.

Why it's recommended:Epoch AI used a self-built experiment to provide quantitative data showing how open-weight models can reproduce vulnerability exploits at low cost during the patch window, and offered defensive recommendations to shorten release cycles.

Also reported by 2 other sources
AGIBOT ZhiyuanModelsAI Score64

Zhiyuan releases the embodied model GO-3 Preview, topping the RoboDojo, RoboTwin, and RoboColiseum leaderboards and planning to open-source it by the end of 2026

On October 29, Zhiyuan released its next-generation embodied foundation model, GO-3 Preview. It topped the three evaluation leaderboards—RoboDojo, RoboTwin, and RoboColiseum—and is planned for full open-source release by the end of 2026.

Why it's recommended:The official announcement provides detailed disclosure of GO-3 Preview’s WRAM architecture, full-stack training methods, and continual-evolution data from real convenience-store scenarios, offering insight into the technical direction of embodied models.

Cursor BlogIndustryAI Score81

Cursor reports that a malicious Anthropic.claude-code 2.1.313 version has appeared in the VS Code extension marketplace

Cursor reported that a maliciously modified version of Anthropic.claude-code 2.1.313 had appeared in its VS Code extension marketplace; it was not developed officially by Anthropic and may have affected up to 4100 users.

Why it's recommended:Cursor provided the number of affected users, detection commands, and a complete remediation script, enabling affected users to check their systems and remove the backdoor.

Reported by 4 other sources
Hacker News: Popular AI postsIndustryAI Score80

Google's market capitalization evaporates by $500 billion as AI Overviews hit search advertising and spark Wall Street concerns

Alphabet's market capitalization fell from $3.97 trillion to $3.46 trillion in a single day as AI search overviews affected advertising revenue, wiping out more than $500 billion. Quarterly total revenue rose 25% to $127.6 billion, and cloud revenue grew 103% to nearly $31 billion, but profit fell 25% and free cash flow was negative by more than $9 billion.

Why it's recommended:The original article breaks down specific figures for AI spending and cash-flow changes in the financial reports of Google, Amazon, and Microsoft, allowing readers to compare the different reactions from the capital markets.

Official account: Moonshot AITutorialsAI Score69

Official Kimi tutorial: How to create 3D works with Kimi K3.1 Pro

Moonshot AI (Kimi) published an official tutorial detailing the 3D creation workflow with Kimi K3.1 Pro. It explains two approaches—parametric modeling and mesh-based modeling—and the corresponding tool choices. The article shares step-by-step instructions for four projects: building an Eastern Heavenly Palace in Unreal Engine, creating a 3D-printable table lamp, generating architecture in Blender through pure code, and building a miniature city with Tripo and Blender. It also provides five usage tips, including that visuals are more powerful than words and that users should build a graybox model first before refining it.

Why it's recommended:The official tutorial publicly reveals the internal 3D workflow for the first time, covering modeling approaches, harness selection, and five reusable tips, allowing readers to get started directly with the examples.

Also reported by 1 other source
PlatformerIndustryAI Score86

Google and Reddit shares plunge on the same day as AI search monetization woes hit the open web

Casey Newton analyzes the market sell-offs Google and Reddit suffered on the same day: Alphabet shares fell 12.6%, wiping out about $500 billion in market value, the second-largest single-day loss in history. Executives acknowledged that AI Mode monetizes less effectively than traditional search, even though AI Mode has about 1 billion monthly users and search revenue still grew 12% to $63.35 billion.

Why it's recommended:The author analyzes Google and Reddit's earnings reports from the same day, noting that AI Mode's weaker monetization than traditional search is prompting websites to add barriers and shrinking the open web.

OpenRouter:AnnouncementsTutorialsAI Score72

OpenRouter Tests Claude Opus vs. Sonnet vs. Haiku 5.5: Which Should You Choose?

OpenRouter tests the three model tiers Claude Opus 5.5, Sonnet 5.5, and Haiku 5.5, recommending Sonnet 5.5 as the default, Haiku 5.5 for high-volume and subagent tasks, and Opus 5.5 for long tasks where errors are costly.

Why it's recommended:The original article tests all three models on the same coding and benchmark tasks, drawing conclusions about cost and reliability when choosing a model for different workloads.

TechCrunch:AIIndustryAI Score78

Jassy Says AWS Will Become a Trillion-Dollar Revenue Business Within Five to Ten Years; 2026 Capex to Reach About $220 Billion

Amazon CEO Andy Jassy said on Thursday's quarterly earnings call that AWS could become a business with $1 trillion in annual revenue within the next five to ten years. Cash capex in 2026 is expected to be about $220 billion, most of it going to AWS, with a substantial increase also expected in 2027.

Why it's recommended:The original article provides AWS's latest financial figures, its roughly $220 billion capex plan for 2026, and Jassy's timeline for reaching $1 trillion in revenue, helping readers assess the pace of cloud providers' AI investments.

Also reported by 1 other source
TechCrunch:AIIndustryAI Score78

Nvidia in Talks to Guarantee Debt Financing for Nscale’s Anthropic Data Center in West Virginia

According to The Information, corroborated by Bloomberg, Nvidia is in talks to guarantee billions of dollars in debt financing for the data center Nscale is building for Anthropic in West Virginia. Anthropic has signed a $44.6 billion contract for the first building.

Why it's recommended:The report examines the structure and risk context of a potential Nvidia guarantee of Nscale's data center debt, helping readers understand interconnected risks in the AI infrastructure financing chain.

Also reported by 1 other source
Load more

Finished reading the news? Let’s talk about your business.

Leave your contact details, and I’ll get in touch with you personally.

WECHAT

Add me on WeChat

Press and hold or select the WeChat ID below to copy it.