United States

Google opened preorders on November 12 for Intelligent Eyewear, screenless glasses built with Samsung, Warby Parker and Gentle Monster that put Gemini in the frame via voice, a 12-megapixel camera and open-ear speakers. Warby Parker frames start at $395 and Gentle Monster's at $449, before prescription lenses, with sales from December 10 for Google accounts in the US and South Korea. Google says camera visuals are not saved to Gemini history or used for training or ads and that bystanders' faces are blurred before processing; Ars Technica notes voice chats get no such guarantee and data is not encrypted on Google's servers.

London-based Nscale's amended IPO prospectus, filed November 12, says it placed large GPU and equipment orders to manage price, allocation and delay-penalty risk under its up to $44.6 billion Anthropic agreements, that some of those commitments cannot be deferred, reduced or cancelled, and that it will need additional financing to meet forecast 2027 obligations. Preliminary, unaudited estimates show a $1.69 billion net loss on $307.5 million of revenue for the nine months to September 30 and total debt of about $5.5 billion, up from $79 million at the end of 2025. No offering size or price range has been set.

A Reuters review of more than 150 Chinese procurement notices, investment plans, contracts and company disclosures identified computing projects with a combined reported value above 23 billion yuan ($3.2 billion) built or planned around Nvidia's restricted H100, H200 and H800 processors, from a planned 6.2 billion yuan centre in Fujian to clusters already running in Wuxi. The total covers buildings, power and other infrastructure, not chip spending alone, and includes plans that may never be completed. Reuters does not establish how the hardware reached Chinese users, and the projects have not been linked to Megaspeed or the allegations around Jensen Huang.

Will DePue of Unemployment Labs published an analysis on November 12 arguing that six models released since April 2026 by DeepSeek, Moonshot AI, MiniMax and Ant Group were initialized on Nvidia Blackwell-class GPUs, which need a U.S. licence for export to China. The method reads leftover starting values in the published weights and matches them to each chip's processor-core layout; earlier models from the same labs read as Hopper. DePue confirmed two matches on physical Blackwell hardware. He says the evidence shows access to the chips, not where they were located, how access was obtained, or whether export controls were violated.

Mirendil said on November 12 that it raised a $1 billion Series A led by Kleiner Perkins and Clearlake Capital, with Battery Ventures, Doerr Capital, Gil & Co, Google, M13, Pear VC and S32 among the investors. The company, which emerged in June, says one researcher working with agents pretrained its Mirendil-Base coding model from scratch in 3.5 months on less than $10 million of compute, and it sells Lor, a private-beta platform that runs hundreds of agents on research work for days. Mirendil did not disclose a valuation, and its training claims have not been independently checked.

Anthropic released self-hosted Claude Managed Agents in private beta on November 12. Customers run the agent harness, sandboxes, credentials and session history in their own cloud or on-premises, operating the server, its database and the workers, and send model requests to Anthropic, Amazon Bedrock, Google's Agent Platform, Microsoft Foundry or their own gateway. Claude itself does not run inside the customer's network. Access is by request through Anthropic account teams, and The Next Web reports that some hosted features, such as webhooks and scheduled runs, are missing.

OpenAI launched Daybreak Red Team and Daybreak Defense Factory on November 12 at its Cyber at the Frontier event. Red Team, in early access, sends coordinated agents that use source code and app context to test web applications, validate reported vulnerabilities and check whether patches hold, within operator-set scope controls. Defense Factory uses agents built on GPT-6 Cyber, according to The Register, to trace vulnerabilities and prepare patches that engineers review before merging. OpenAI published no pricing or independent performance results for either product.

Google released Gemini 4 Hydrogen on November 12, a model it pitches as the fast execution engine for coding and agent work beside the larger Gemini 4 Carbon and Argon. It costs $1.50 per million input tokens and $7.50 per million output tokens across a 1 million-token context and is rolling out in the Gemini API, AI Studio, the Gemini app, Search and Antigravity. Artificial Analysis scores it 51 on its Intelligence Index at the default high setting and 54 at xhigh, where it ranks 16th of 78 models; most other benchmark figures are Google's own.

Irreplaceable Education Fund, Public Citizen, Heat Initiative and The Alliance for Secure AI filed suit on November 12 in San Francisco Superior Court against Anthropic, Google, Meta, OpenAI and xAI. Invoking California's Unfair Competition Law and a public-nuisance theory, they ask the court to bar the companies from developing or deploying AI systems more capable than their current most advanced models. The complaint leans on the companies' own warnings about catastrophic risk and on past incidents involving AI agents. The claims are allegations, and the defendants had not responded when the suit was reported.

Databricks on November 11 announced a second-generation Genie, an enterprise agent that it says runs in the background on its own compute, delegates to sub-agents, tracks long-running work and brings decisions to people through a priority inbox and email. Every action is grounded in Genie Ontology, a model of a company's definitions, metrics, people and processes, and governed by Unity Catalog and Unity Gateway. Genie and Genie Ontology are generally available; the always-on Genie, Genie Search and Ontology Manager are coming soon to preview and beta programs.

United Airlines said on November 11 it is restricting unauthorized third-party AI agents from its website, app and booking channels, including indirect use through travel agencies and aggregators, Reuters reported. United's policy page says agents may still help with research but may not search, book, buy, manage or check in on its platforms without permission, and that automated check-in could breach FAA and DOT rules. Bookings made by agents before November 11 will not be cancelled. JetBlue's president said airlines should follow customers' lead.

Chinese AI companies are trying to rent Nvidia-powered computing capacity at data centres outside China, three people familiar with the matter told Reuters on November 12. They fear the allegations around former Nvidia chief executive Jensen Huang could add momentum to the Remote Access Security Act, a House-passed bill that would let the Commerce Department regulate cloud access to restricted chips. A major Chinese cloud provider has discussed locking in overseas capacity and smaller labs are approaching Southeast Asian rental providers, the people said. The companies were not named.

Runway released Praxis-1 on November 11 with open weights under Apache 2.0, including inference and fine-tuning code. The 27-billion-parameter model predicts video and robot actions together and scores 46.2% success on Nvidia's RoboLab-120 simulation benchmark, first among world action models and ahead of FLUX 3 Action (42.9%). On the leaderboard's overall table it ranks fifth, behind systems including an agentic language-model policy at 86.0%. Runway says it adapts to new robots with light fine-tuning.

Torus, a Salt Lake City maker of battery and flywheel storage, announced $500 million in growth capital led by Magnetar on November 11, which Bloomberg described as a Series C, and launched Foundry, prefabricated units that deliver 1 to 20 megawatts of AI inference computing behind the meter at existing commercial and industrial sites. Torus says a site can go from contract to running workloads in 12 to 16 weeks and plans more than 47,000 GPUs in Foundry units by the end of 2027. No valuation was disclosed.

Santosh Janardhan, Meta's head of infrastructure and co-head of engineering since joining in 2009, told staff in a memo on November 11 that he will step down, CNBC and TechCrunch report. Dave Brown, who spent 19 years at Amazon and joined Meta in July, will lead infrastructure within the Meta Compute initiative. Janardhan will stay through the first quarter of 2027 for the transition and then move to an advisory role on Meta Compute. Meta did not give a reason.

Northwestern Polytechnical University, a Chinese defence-linked university on the US Entity List since 2001, bought two Nvidia H200 chips in October for 874,700 yuan from a domestic supplier, according to procurement records reported by Reuters on November 11. Its cybersecurity school also rented access to servers with H200 chips this year. A Reuters review of 26 records found other Chinese universities buying or renting restricted Nvidia chips through local suppliers. Nvidia, the university and the US Commerce Department did not respond to Reuters.

Google launched the Gemini Decisions API on November 11, powered by a new model, Gemini Decision 1, that scores closed-set questions (yes/no, 2 to 255 choices, or rubrics) against one text, image, audio, video or PDF input and returns probabilities and confidence scores instead of generated text. It handles up to 1,000 questions per request at $0.10 per million input tokens with no output charge. Google's own benchmarks put it ahead on five of seven shared tests; the independent Artificial Analysis Decision Index ranks it ninth of 18, per VentureBeat.

AWS released AWS Emulator on November 11, a free local emulator built from more than 400 API models covering more than 200 AWS services, so developers and coding agents can test against AWS APIs without an account or network calls. AWS says autonomous agents wrote all of the code and its team has not written or reviewed any of it, relying instead on agent-run tests checked against real AWS and an automated merge gate. The code is public under CC0; unsupported operations return 501 Not Implemented.

Security firm Depi disclosed on November 11 a chain that started with a prompt injection in a GitHub Actions workflow on OpenAI's Codex repository, where any outside user could trigger the Codex agent by opening an issue. Depi says it escaped the agent's sandbox, took workflow tokens, poisoned the repository's build cache and could have obtained the credentials needed to publish a backdoored @openai/codex package on npm. OpenAI fixed the core bug and paid an initial $1,000 bounty with the report still under review, Depi says. No malicious release has been reported.

Oura agreed on November 10 to acquire Counsel Health, an AI-native primary care company that combines medical AI with in-house physicians, for $625 million in cash subject to adjustments. Counsel's service has run inside the Oura app since June; members chat with its AI and can escalate to a physician who can diagnose, prescribe and refer. Oura reported preliminary, unaudited fiscal fourth-quarter revenue of $533 million, up 153%. The deal is expected to close by year-end, subject to conditions.

Treasury Secretary Scott Bessent said on November 10 that next week's Trump-Xi meeting will advance a U.S.-China dialogue on 'superintelligence,' in which the two sides expect to exchange best practices for evaluating and testing advanced models before release and have agreed to a direct channel for alerting each other to serious incidents. Reuters and CNBC report that he made no commitment to limit AI development. No thresholds, agencies or procedures for the channel have been published.

Gov. Kathy Hochul launched the New York Center for AI Responsibility and Research at SUNY Binghamton on November 10, pledging independent testing of frontier AI systems with more than 80 researchers and Empire AI computing. The state's new DIGIT office also opened the portal large frontier AI developers must use to register under the RAISE Act by January 1, 2027, when safety-framework, incident-reporting and transparency duties begin. The announcement does not explain how the center will obtain access to developers' models.

Railway's incident report says an attacker who escaped its build isolation on October 20 decrypted stored environment variables for at most 1.85% of workspaces, about 100,000, and specifically searched for AI provider keys. The company estimates unauthorized spend on the stolen keys at about $740,000 and says it will cover charges that AI providers will not refund. It believes the keys went to resellers for use in model distillation but has not named the attacker or said how the estimate was calculated.

OpenAI launched Businesses in ChatGPT in beta on November 10. Through Business Hub, eligible companies can claim and verify their profile, update details and product feeds, and see aggregated mentions, website clicks and discovery topics. Eligible U.S. ecommerce and online-service businesses can also build Business Agents that users can chat with in a labeled, separate conversation, and can sponsor those agents as destinations for ChatGPT Ads. OpenAI did not detail its verification methods.