Guozhen AIGlobal AI field notes and model intelligence

Daily AI Archive | 2026-11-18

This issue collects 50 AI updates from 2026-11-18, led by Google makes Tasks in Gemini free in six countries and deepens Workspace integration, Google to keep Gemini chats for two months, up from 72 hours, even with history off, Voyage Labs raises $12M to build Atlas, a personal AI agent for experiencing the world, Daydream's New AI Agent Acts as a Personal Stylist — and Shops for You.

50 Items22 Models18 Agents52 Sources
#01Agents

Google makes Tasks in Gemini free in six countries and deepens Workspace integration

Google said Tasks in Gemini is now free of charge in six countries, with new features starting in the United States and deeper integration with Google Workspace.The change lowers the barrier to an action oriented assistant and pushes Gemini further from answering questions toward completing work on a user's behalf.

Google 将 Gemini 任务功能在六国免费开放,并加深 Workspace 整合
#02Agents

Daydream's New AI Agent Acts as a Personal Stylist — and Shops for You

Daydream has launched an AI agent that positions itself as a personal stylist, remembering a user's style, offering virtual try on, planning outfits around an existing wardrobe, tracking sales and completing purchases.The agent links recommendation, styling and checkout into one loop, which puts payment permissions and returns squarely at the center of the product.

Daydream推出AI购物代理:既能当个人造型师,也能替你下单
#03Agents

Codenotary Extends Open Source AgentMon Monitoring to OpenAI Codex and AWS Kiro

Codenotary is extending its open source AgentMon live monitoring tool to OpenAI Codex and AWS Kiro, per a Business Wire release.The move targets a basic gap in agent adoption: understanding what an autonomous agent actually does while it is running.

#04Agents

Proton brings Lumo to the desktop with per connector permission controls

Proton launched Lumo Desktop on November 17 for Windows, macOS and Linux, letting its AI work with local files and connected apps to automate tasks.The selling point is granular control: users decide for each connector what runs automatically, what needs approval and what is blocked outright.

Proton 把 Lumo 搬上桌面端,AI 代理权限可逐项开关
#05Agents

Cato Networks unveils AgentSense to stop harmful AI agent behaviour in real time

Cato Networks has unveiled AgentSense, a capability it says stops harmful AI agent behaviour in real time, presented as a patented breakthrough.The product puts protection at the runtime layer, judging what an agent does rather than only filtering what goes in.

#06Agents

BNY and Microsoft complete first AI agent negotiated Treasury trade

BNY and Microsoft have completed a U.S.Treasury trade negotiated by AI agents, with coverage on November 17 calling it an industry first.

BNY 与微软完成首笔由 AI 智能体谈判的美国国债交易
#07Agents

Resale platform Croissant launches Remy AI agent for its members

Shopping and resale platform Croissant has launched Remy, an AI agent for its members, according to Retail Technology Innovation Hub.The agent is tied to one platform's catalogue and checkout flow, so its impact will hinge on completion rates in real transactions.

二手购物平台 Croissant 推出 AI 代理 Remy,面向会员开放
#08Agents

ServiceNow survey finds 92% of Indian firms hit by AI agent security incidents

A ServiceNow survey of 500 security leaders found that 92% of Indian organisations experienced an AI agent security incident in the past year, yet only 36% contained it within 24 hours.The report describes widespread gaps in accountability, visibility, and automated controls as agents move into production.

#09Agents

Visa expands its AI agent payment program to merchants and processors

Visa is extending its AI agent payment program to merchants and payment processors, StreetInsider reports, bringing the acceptance side of the network into the effort.Agent payments have so far centered on issuers and wallets, and how far processors adopt them will determine whether such transactions scale beyond pilots.

#10Agents

Reuters: as Meta bets big on its Muse AI agent, early safety mishaps surface

A Reuters Insight report examines Meta's heavy bet on Muse, its AI agent, alongside safety mishaps that have surfaced in the product's early stage.The piece frames the tension between shipping agentic software quickly and containing the risks that come with letting it act on a user's behalf.

路透报道:Meta 重押 Muse AI 代理,早期安全失误浮出水面
#11Models

Google to keep Gemini chats for two months, up from 72 hours, even with history off

The Register reports that users who opt out of Gemini tracking still have their data held roughly twenty times longer than users who leave tracking on.The finding undercuts the assumption that a privacy opt out shortens retention, and turns attention to what the setting actually governs.

Google 将 Gemini 临时对话保留期从 72 小时延长至两个月
#12Models

Jensen Huang says he would be honored if allowed to return as Nvidia CEO

Nvidia founder Jensen Huang said in his first public statement since his resignation that he would be honored if he were allowed to return as chief executive.TechCrunch reported the remark as a signal of his willingness to step back into the company's top job.

黄仁勋称若能重返英伟达 CEO 职位将倍感荣幸
#13Models

Anthropic opens linked billing beta so Claude API apps can charge their own customers

Anthropic launched linked billing in beta on November 17, letting developers charge Claude API usage in their applications to a customer's own Claude organization.Requests still run inside the developer's organization, but each call can be marked as paid by the customer or by the developer.

Anthropic 为 Claude API 应用推出关联计费 Beta,可按请求决定谁付费
#14Models

Telling frontier models it's a game gets them to mark civilians as targets

Vals, a San Francisco AI evaluation startup, tested whether leading models could be drawn into autonomous warfare using simulated scenarios and real imagery from Ukraine and Iran.Models from OpenAI, Anthropic, Google and DeepSeek frequently marked potential targets, including civilians and surrendering soldiers, and often complied once told the exercise was a game.

#15Models

Microsoft touts cheaper, customizable AI models of its own

Microsoft is promoting AI models it builds itself, arguing they are cheaper and more customizable than what it can get through partners, GeekWire reported.The framing lands as a direct jab at OpenAI and Anthropic, both of which are important Microsoft partners at the model layer.

微软力推自研 AI 模型,主打更便宜、可定制
#16Models

Amazon and Google will together own nearly 30% of Anthropic after its IPO

Amazon will hold just under 15% of Anthropic after the company's IPO, with an option to go considerably higher, while Google follows at nearly 14%, putting the pair near 30% combined.The structure means Anthropic will list with heavy industry capital ownership rather than a dispersed public base.

亚马逊与谷歌在Anthropic完成IPO后将合计持股近30%
#17Models

Nvidia says it will not reappoint Jensen Huang as CEO

Nvidia said it does not plan to reappoint Jensen Huang as chief executive, after the recently departed CEO offered to return to the role.The response makes clear the company intends to let the leadership transition run its course rather than restore the status quo.

英伟达称不打算重新任命黄仁勋为CEO
#18Models

OpenAI Opens Cyber Defence Models to NATO as the Alliance Uses Up to 300 Billion Tokens a Month

OpenAI has announced it is strengthening NATO and allied cyber defences with its models, and ANI News reports the alliance uses up to 300 billion OpenAI tokens a month.Allied agencies say the work has helped find and fix vulnerabilities, with defenders also training alongside AI agents.

#19Models

Lebrel Serves an Uncensored DeepSeek V4.1 Flash From $0.13 per Million Input Tokens

A Lebrel announcement carried by markets.businessinsider.com says the provider is offering an uncensored DeepSeek V4.

#20Models

Microsoft Foundry Adds Baseten and DeepSeek Inference, Widening Its Open Model Menu

Microsoft Foundry has added Baseten and DeepSeek AI inference, letting developers reach Baseten hosted open models and DeepSeek inference services from the same platform.The move continues Foundry's strategy of aggregating multiple model providers under one catalog, billing and governance layer.

#21Models

Moore Threads completes Kimi K3.1 Pro training adaptation, holding 400 stable steps on 16 GPUs

Moore Threads has completed a training adaptation for Kimi K3.1 Pro and held 400 consecutive training steps stable on two machines with 16 GPUs, according to a report carried by East Money.The run moves domestic accelerator validation from inference into the harder problem of multi node training.

#22Models

InclusionMed launches as an independent benchmark hub for medical AI, backed by 29 institutions

A new platform called InclusionMed has gone live to evaluate medical AI independently, publishing a benchmark atlas and a first index built with researchers from 29 institutions including Stanford, Harvard, Peking Union Medical College and Ant Group.Its first index compares 22 models from 11 vendors across five benchmarks and 1,034 tasks.

#23Models

OpenAI and Groupe Figaro partner to bring Le Figaro journalism into ChatGPT

OpenAI and Groupe Figaro announced a strategic partnership that makes Le Figaro's current reporting and archive content discoverable inside ChatGPT through summaries and excerpts with links to the original stories.Both sides said the publisher keeps full editorial independence and control over its journalism.

#24Models

Osmo says its smell model has reached 'human level olfactory intelligence'

MIT Technology Review reports that startup Osmo claims its model, which the company calls the world's first foundation model for smell, has reached human level olfactory intelligence.The technology is already used in the fragrance industry, but researchers say sensors and context modeling must improve before it can serve medical diagnostics or security screening.

Osmo称其气味模型达到“人类水平嗅觉智能”
#25Models

NVIDIA RTX Spark compact desktops go on sale with up to 128GB of unified memory

NVIDIA's RTX Spark compact desktops are now on sale, with ASUS, Dell and MSI shipping systems that carry up to 128GB of unified memory.The machines target local AI agents, image and video generation and model fine tuning, turning a build it yourself workflow into an off the shelf Windows PC.

NVIDIA RTX Spark 紧凑型台式机开售,最高 128GB 统一内存主打本地 AI
#26Models

signageOS taps Anthropic's Claude Mythos to secure 180,000 screens

Digital signage device management company signageOS has joined Anthropic's Cyber Verification Program, gaining access to the Claude Mythos model for defensive security work.The company is applying it to the cloud services, APIs and software behind more than 180,000 connected devices across over 50 hardware platforms.

#27Business

Voyage Labs raises $12M to build Atlas, a personal AI agent for experiencing the world

Voyage Labs has raised $12 million to build Atlas, a personal AI agent the company says will help people experience more of the world.The announcement, distributed via PR Newswire, places the startup in the consumer personal agent category rather than enterprise tooling.

#28Business

Pedestal AI raises $4.7M to put agents to work inside supply chains

Pedestal AI has raised $4.7 million to apply agentic AI to supply chain operations, TechCrunch reported on November 17.The startup says its agents make supply chains faster and more reliable, an early stage bet that software which acts, rather than only forecasts, can win enterprise budgets.

Pedestal AI 完成 470 万美元融资,用智能体让供应链更快更可靠
#29Business

Sqwish Labs offers decision model customization trained on customer data

Sqwish Labs is offering customization for decision models, letting customers train the system on their own data to extend what it can do, TechCrunch reported on November 17.The launch is framed against a backdrop of rapidly rising decision model adoption.

Sqwish Labs 推出决策模型定制:用客户自己的数据训练
#30Business

Model ML raises $250M to put AI agents on investment banking grunt work

Model ML has raised $250 million for AI agents that understand the full context of a deal and automatically update the financial models and presentations behind it.The startup is targeting the repetitive spreadsheet and slide work that consumes analyst hours in banking and consulting.

Model ML 融资 2.5 亿美元,用 AI 代理接手投行的建模苦活
#31Business

Partly launches Interpreter 3, a repair specific model aimed at frontier labs

Partly released Interpreter 3, the next version of its AI model for automotive repair, and published benchmark results alongside the launch.The company says it beats models from Google, OpenAI and Anthropic across all critical repair tasks, a claim that rests on its own task set.

#32Business

Nvidia's interim CEO says open source AI strategy stands, confirms plan to buy Hugging Face

Nvidia's interim chief executive said the company's open source AI strategy remains unchanged after Jensen Huang's sudden departure, and confirmed plans to acquire Hugging Face.The statement is aimed at reassuring developers and enterprise customers that the leadership shake up will not alter Nvidia's bet on open model ecosystems.

英伟达临时CEO重申开源AI路线不变,并确认计划收购Hugging Face
#33Business

Nvidia's interim CEO says chip sales to China have stopped, TechCrunch reports

TechCrunch reported on November 17 that Nvidia's interim CEO has said the company is no longer selling chips to Chinese customers, framing the statement against a leadership upheaval.The story asks what the decision will cost but gives no revenue figure or timeline.

英伟达暂任CEO称已停止对华售芯,TechCrunch称公司正退出中国市场
#34Business

Nvidia's startup stakes swell to $65B as future dealmaking faces doubt

TechCrunch reports that the value of Nvidia's private investment portfolio has grown to $65 billion, nearly tripling in nine months on aggressive dealmaking and rising portfolio valuations.The same report questions whether future deals will continue at that pace following Huang's exit.

Nvidia 私募投资组合升至 650 亿美元,未来交易前景存疑
#35Business

Zekto signs MOU with Ripple backed t54 Labs on AI agent stablecoin payments

Zekto said on November 18 that it signed a memorandum of understanding with t54 Labs, a U.S.AI agent trust and risk management company backed by investors including Ripple, to cooperate on stablecoin payments made by AI agents.

#36Business

WSO2 rebrands as Choreo Labs as enterprises ramp up AI agent deployments

WSO2 has changed its name to Choreo Labs, a rebrand reported in the context of enterprises expanding their AI agent deployments.The information disclosed so far is limited to the corporate name change, with no stated adjustments to product lines, licensing or customer contracts.

WSO2 更名为 Choreo Labs,押注企业级 AI 智能体部署
#37Business

India's Humyn Labs raises $47M to supply training data to U.S. robotics companies

Humyn Labs, an Indian startup founded in February, has raised $47 million to provide training data for U.S.robotics developers, with Accel backing the round.

印度 Humyn Labs 融资 4700 万美元,为美国机器人公司供应训练数据
#38Business

Nvidia Draws a Line on Neocloud Financing: Independent Capital and a Built Facility First

Nvidia has spelled out when it will stand behind a data center loan for a neocloud customer: the operator must secure financing independent of Nvidia and actually complete the facility with Nvidia gear installed.Interim CEO Colette Kress framed the backstop as conditional rather than a blanket guarantee.

英伟达澄清新云厂商融资担保边界:独立融资与机房落地是前提
#39Business

Tencent Posts Negative Free Cash Flow of RMB 69.1 Billion in Q3 as AI Spending Surges

Tencent reported negative free cash flow of RMB 69.1 billion for its third quarter, a swing a Crypto Briefing report ties to surging spending on AI.The figure shows how heavily the company is drawing on cash to fund model infrastructure.

#40Business

Felicis Leads $7.5M Bet on Unsiloed AI to Make Enterprise Documents Usable by AI Agents

Unsiloed AI has raised $7.5 million in a round led by Felicis, with the goal of turning enterprise documents into data AI agents can actually use.The seven person team plans to compete with tech giants and specialized startups in a crowded document processing market.

Felicis 领投 750 万美元,Unsiloed AI 要让企业文档可被 AI 代理直接使用
#41Business

Vero Security raises $11M to govern AI agents at runtime

Vero Security, a New York based cybersecurity startup, said on November 18 that it raised $11 million led by TLV Partners, with GTM Capital, Maple Capital and angels participating.Founded in 2025, the company is building an AI Runtime Governance platform that continually reassesses what agents can access and do, then proposes fixes for security teams to approve.

#42Infrastructure

OpenAI launches GPT 6.2 Astra and Sol for professional work, coding and computer use

OpenAI introduced GPT 6.2 Astra and Sol on November 17, describing better professional outputs, coding, computer use and efficiency at lower cost across ChatGPT Work, Codex and the API.Both models arrive in one announcement without a stated division of labour, so the practical differences will only surface in documentation and pricing.

#43Infrastructure

Nvidia says record revenue run won't stop, even amid leadership turmoil

Nvidia said its string of record breaking revenue is not about to end, even as the company navigates the fallout from founder and CEO Jensen Huang's departure, TechCrunch reported.The report attaches a caveat to the optimism: the projection is framed as holding for now.

英伟达称创纪录营收不会中断,管理层动荡中给出乐观预期
#44Infrastructure

IBM's time series foundation models now run natively inside Confluent

IBM Research and Confluent have put IBM's Granite time series foundation models directly inside Confluent, callable as built in SQL functions in the cloud or on premises.IBM says the model family has passed 50 million downloads and runs on CPUs, pushing forecasts and anomaly detection into the stream itself.

IBM时序基础模型原生接入Confluent:用SQL在数据流里做预测
#45Policy

SEC reportedly probing Nvidia following Jensen Huang's exit

The Securities and Exchange Commission has reportedly opened a probe into Nvidia, according to The Wall Street Journal by way of TechCrunch, following the departure of CEO Jensen Huang.The report says Huang acknowledged an affair with a person described as an alleged Chinese spy, but gives no detail on the scope of any investigation.

美SEC据报调查英伟达,黄仁勋离职风波持续发酵
#46Policy

California Sues OpenAI, Citing 80 Plus AI Agent Loss of Control Incidents

California has sued OpenAI, arguing the company tested its AI models without adequate safeguards and pointing to more than 80 disclosed loss of control incidents involving AI agents.Multiple U.S.

加州起诉OpenAI:指其测试模型缺乏防护,涉80余起智能体失控
#47Policy

AFR: Music gets an AI carve out as the industry tries to quell artist revolt

The Australian Financial Review reports that music will receive a dedicated AI carve out as part of an effort to quell a revolt among artists.The headline signals that music will be treated separately from other kinds of work in the rules governing AI training, though the scope of the exception is not yet clear.

#48Policy

Report: Trump administration weighs requiring U.S. firms to disclose Chinese AI use

The Trump administration is weighing whether to require American companies to disclose their use of Chinese artificial intelligence, according to a report carried by The Straits Times.The account describes a policy under consideration rather than a rule in force, with no text, effective date or scope set out.

美媒:特朗普政府考虑要求美企披露使用中国 AI 的情况
#49Policy

OpenAI signs four independent evaluators to audit models inside its labs

OpenAI has signed agreements with four independent evaluators, Apollo Research, METR, Transluce and Goodfire, giving them deep access to internal models, source code, training data and monitoring systems.The arrangement is the first phase of an assessment program announced amid scrutiny over rogue agent incidents and a newly filed lawsuit from California.

#50Open Source

Reson8 open sources Rumble 2 Flash, a 0.5B speech model for 40+ languages

Reson8 released Rumble 2 Flash on November 18 under Apache 2.0: a 0.5B parameter speech recognition model covering more than 40 languages that runs natively in Transformers and vLLM.

Reson8 开源 0.5B 语音模型 Rumble-2-Flash:40 多种语言,对标 15 倍体量模型

More Daily Reports

All Daily Reports