The Wire
Daily AI news, learning & projects
News
Timeline
Models
Learning
GitHub
?
Latest
Daily
Social
Videos
‹ Tuesday's edition
·
Daily edition
·
Last 24 hours ›
Wednesday 11 November
All
OpenAI
Anthropic
Google
Other
AI industry
15:00
Google DeepMind launches AlphaGenome 2 to predict DNA effects in specific cells
Google DeepMind has released AlphaGenome 2, an AI model that predicts how DNA variants affect individual human cell types and states. Trained on over 100 million cell profiles across 68 tissues, it nearly doubles the number of gene-expression variants found compared to its predecessor. It is available now for non-commercial research.
3 sources
Techmeme
Google DeepMind
The Verge
Nvidia
Developing · since Thu
Nvidia board reportedly prepares to oust CEO Jensen Huang
NEW
The New York Times reports Huang also had an extramarital relationship with Nvidia VP of healthcare Kimberly Powell, and the Wall Street Journal says the board may meet Wednesday to oust him.
Nvidia's board is reportedly meeting as soon as Wednesday to decide Jensen Huang's future. The move follows revelations of his affairs with an alleged Chinese spy and an Nvidia executive, and Congress has summoned him for a briefing. Nvidia shares fell more than 5% Tuesday.
7 new · 12 sources
TechCrunch
The Verge
twitter.com
Techmeme
The Wall Street Journal
The New York Times
Wired
BBC News
+4 more
Policy & law
18:40
United Airlines will block AI agents from booking flights
United Airlines says personal AI assistants will not be allowed to complete flight bookings, citing data-privacy risks and potential FAA and DOT rule violations. CEO Scott Kirby said he does not want another middleman. The move limits emerging agents like Meta's Muse and OpenAI's Dots.
3 sources
The Verge
cbsnews.com
Techmeme
AI industry
13:20
Fractile raises $720M for AI chips at $7B valuation
London-based AI chip startup Fractile raised $720 million from Thrive Capital, Founders Fund and other investors, sources told the Financial Times. The round values the company at about $7 billion post-money, up from roughly $1.1 billion in May, making it one of the best-funded players building purpose-built semiconductors for AI workloads.
2 sources
TechCrunch
Techmeme
Google
18:00
Google launches Gemini Decisions API with confidence scores
Google introduced the Gemini Decisions API, powered by a new Gemini Decision 1 model, which returns predefined answers with confidence scores for up to 1,000 yes/no or multiple-choice questions. Priced at $0.10 per 1M input tokens, it targets fast, calibrated multimodal decision tasks rather than open-ended generation.
2 sources
Techmeme
Google
AI industry
14:00
Rox raises $100M Series B at $1.3B valuation
Rox, a startup building AI agents for customer research and outreach, raised a $100 million Series B co-led by General Catalyst and March, valuing it at $1.3 billion. The funding highlights investor interest in AI agents automating enterprise sales workflows.
2 sources
Techmeme
TechCrunch
Policy & law
07:00
England regulator urges return to AI-free university assessments
England's higher education regulator, the Office for Students, said AI has made assessing coursework close to impossible and that degree credibility is under threat. It recommended universities use traditional exams and in-person presentations. The advice responds to lecturers' concerns about verifying student work.
2 sources
Techmeme
The Guardian
Google
16:00
Google Labs launches Tabletop AI collaboration experiment
Google announced Tabletop, a new Labs experiment that uses AI to generate customizable collaborative spaces for groups. Users can describe what they want and get interactive tools like itineraries, polls and brackets for planning with family or teams. It is currently waitlist-only.
2 sources
Google
The Verge
Policy & law
06:15
Commerce finalizing rule to block remote AI chip access from China
The Commerce Department is finalizing a semiconductor rule to stop Chinese AI firms remotely accessing advanced chips via data centers and curb smuggling. It follows Reuters reporting that a PLA-linked Chinese university obtained Nvidia H200 chips despite US export restrictions.
1 source
Techmeme
Nvidia
03:00
Nvidia uses vision AI and robotics to enhance waste recycling efficiency
Nvidia is utilizing vision AI and robotic systems to process mixed waste materials. This technology identifies and isolates recoverable items on conveyor belts, helping divert substantial tons of material from landfills.
1 source
Nvidia
AI industry
11:50
SoftBank reserved 13GW of gas turbines for AI power, then took a $3.7B hit
SoftBank reserved over 13GW of gas turbines, roughly the output of six Hoover Dams, to power AI data centers, according to the Wall Street Journal. The company later canceled many of those reservations, taking a $3.7B earnings hit mostly from the cancellations. The episode underscores the financial risk of securing massive energy capacity for AI.
1 source
Techmeme
AI industry
Developing · since Tue
Indie games festival in turmoil after Big Walk developer confirms AI use
House House has confirmed it used AI to fix a bug in Big Walk, saying under 0.1% of the game's code came from LLM suggestions. Many Independent Games Festival judges say any AI use disqualifies a game, prompting a broader industry debate.
2 new · 3 sources
Ars Technica
Techmeme
The Verge
Anthropic
Developing · since Tue
Anthropic IPO oversubscribed as bankers weigh raising price toward $1.5T
NEW
The IPO is oversubscribed many times over, bankers may raise the price toward a $1.5 trillion valuation, an Australian retail offer opened, and Goldman projects $600B+ revenue by 2030.
Demand for Anthropic's IPO has exceeded available shares several times over, and bankers may lift the price range, pushing its valuation toward $1.5 trillion or more. An Australian retail offering opened, and Goldman projects revenues above $600 billion by 2030.
1 new · 6 sources
Techmeme
TechCrunch
The Guardian
The Verge
BBC News
Anthropic
AI industry
Developing · since Tue
Ars Technica: Signal encryption passes AI-powered formal verification checkup
NEW
DeepMind plans to open-source AlphaProof Code, and Ars Technica notes the four-week project suggests AI could make formal verification far more practical.
Coverage this week highlights Google DeepMind's four-week project using AlphaProof Code to formally verify Signal's core encryption, with agents producing three million lines of Lean proofs. DeepMind plans to open-source the tool, and analysts see AI making formal verification far more practical.
1 new · 5 sources
Ars Technica
ZDNET
Google DeepMind
Techmeme
Wired
OpenAI
Developing · since Tue
OpenAI quarterly revenue reportedly tops $10 billion for first time
OpenAI told investors its third-quarter revenue jumped roughly 50% to about $10 billion, a rebound from 17% growth in Q2, according to leaked documents and media reports. Its adjusted operating margin improved from -150% to -109%, though the company remains far from profitability.
1 new · 2 sources
TechCrunch
Techmeme
AI industry
Developing · since Tue
TSMC to invest $1 billion in Anthropic and OpenAI
NEW
The $1 billion total and Anthropic stake at IPO price came amid a wobble in AI-related trade, underscoring the deal's timing as a bet against market nerves.
TSMC plans to invest up to $1 billion across Anthropic and OpenAI, putting up to $500 million into each. The chipmaker's move, reported amid AI trade wobbles, signals confidence in long-term AI demand and adds to the boom's circular financing.
1 new · 2 sources
Ars Technica
Techmeme
Nvidia
Developing · since Mon
Megaspeed founder back in Beijing as chip export fight roils Washington
NEW
Huang is reportedly back in Beijing with security officials aware; Trump approved curbs Nov. 5 and Bessent confronted Lutnick over the Entity List omission, which Commerce denies.
Megaspeed founder Alice Huang has returned to Beijing with the knowledge of security officials, Bloomberg reports. Trump approved curbs on suspected Nvidia chip smugglers Nov. 5, though Commerce Secretary Lutnick has stalled them, and Bessent confronted him over Megaspeed's exclusion from the Entity List.
1 new · 1 source
Techmeme
Anthropic
Developing · since Tue
Anthropic launches Claude Meetings as AI office suite expands
Anthropic has unveiled Claude Meetings, letting Claude join video calls to draft plans and tasks in real time, alongside Claude Whiteboards and Claude Sheets. The move follows OpenAI's workplace tools push and positions Anthropic to compete with Microsoft and Google. Claude Data is now generally available.
1 new · 6 sources
ZDNET
The Register
claude.com
Techmeme
TechCrunch
The Verge
More from the day
AI industry
12:21
ByteDance plans Muse-like AI assistant launch within weeks
Bloomberg reports, citing sources, that ByteDance co-founder Zhang Yiming said the company plans to launch a Muse-like AI assistant in the coming weeks. He also said ByteDance aims to reach the AI frontier by 2029-2030, building on its lead in AI video generation.
1 source
Techmeme
AI industry
15:00
Databricks introduces AI agent for unvarnished employee feedback
Databricks has launched its Genie AI agent to assist CEO Ali Ghodsi. This tool is designed to gather candid information from the company's 12,000 staff members. The system aims to uncover internal issues that employees might not report directly.
1 source
TechCrunch
AI industry
20:25
AWS launches Emulator tool written entirely by AI agents
Amazon Web Services released AWS Emulator, a free tool that simulates over 200 AWS services locally. The company says AI agents wrote the entire codebase and that no engineers wrote or reviewed any line of it, presenting the fully AI-built approach as a selling point for developers.
1 source
Techmeme
AI industry
23:50
CoreWeave plans $4B+ investment-grade loan as soon as next week
CoreWeave intends to raise at least $4 billion through an investment-grade loan, potentially priced as soon as next week. The AI cloud computing provider expects yields of 7%–7.5%, roughly half what it pays on its junk bonds, reflecting improving credit access to fund its AI infrastructure buildout.
1 source
Techmeme
AI industry
12:30
German defense firm Helsing launches AI intelligence platform Clarity
Helsing introduced Clarity, an intelligence platform built around AI agents customized for military use. It positions the German defense company against rivals Palantir with Maven and Anduril with Lattice. The launch signals growing competition in AI-powered defense intelligence systems.
1 source
Techmeme
Nvidia
00:25
AMD's Zen 6 EPYC Venice details emerge: TSMC N2P and strong performance gains
Tom's Hardware published a deep dive into AMD's Zen 6 EPYC Venice server CPUs, the first chips built on TSMC's 2nm N2P process. The 9686F reportedly often outperforms Nvidia's Vera, and dual 9996 chips showed up to 93% gains over dual Zen 5 9965 in tests.
1 source
Techmeme
Google
11:00
Parents say Gemini on school laptop failed to flag teen's suicide risk
A 13-year-old girl used Google's Gemini on her school Chromebook to plan her suicide, and the school's monitoring systems did not flag it for weeks, according to her parents. The case raises questions about AI chatbot safeguards on school devices and the limits of student monitoring tools.
1 source
Wired
AI industry
12:00
AI datacentres plan 181GW of on-site gas power, research finds
Research reveals AI datacentre developers in the US, UK and Europe plan on-site gas power stations totalling 181GW of capacity. The buildout is driven by delays connecting projects to national grids, and comes despite climate concerns about fossil fuel use.
1 source
The Guardian
AI industry
15:21
Edelbyte raises $70 million to build data analytics platform for AI agents
Edelbyte, a startup based in Tel Aviv and San Francisco, has emerged from stealth after securing $70 million in funding. The company is developing a specialized data analytics platform designed to support and optimize AI agents.
1 source
Techmeme
Google
12:05
Google AI agents find unmapped Maya pyramid in NASA lidar data
Google software engineers deployed 48 AI agents to analyze NASA lidar data covering Maya regions in Mexico. The agents reportedly identified an unmapped pyramid within 72 hours. The result shows how AI agent systems can accelerate archaeological discovery in large scientific datasets.
1 source
pietromiotti.com
Nvidia
23:50
TechCrunch publishes timeline of alleged Jensen Huang scandal
TechCrunch has published a timeline article describing what it calls an unfolding scandal involving Nvidia CEO Jensen Huang. The outlet reports the affair could have ramifications for national security, given the ongoing US-China race to build advanced AI infrastructure.
2 sources
Techmeme
TechCrunch
Policy & law
20:02
Data center boom hits siting, permitting and power hurdles
TechCrunch reports that many announced AI data center projects may never be built, with billions in planned infrastructure facing real-world constraints. Local opposition, permitting delays and a scramble for electricity are slowing the buildout even before construction begins. The reports suggest the actual pace of AI infrastructure expansion may fall well short of headline announcements.
1 source
TechCrunch
AI industry
14:00
Adobe unveils experimental AI tools for photos at Max
Adobe announced experimental AI features at its Max conference, including tools that turn images into explorable 3D scenes, let users design AI characters through conversation, and make photo editing work like copy and paste. The projects are research previews rather than shipping features, showing directions Adobe may take its creative AI tools.
1 source
The Verge
AI industry
13:25
Energy storage startup Torus raises $500M for container-sized AI data centers
Torus, a Salt Lake City energy-storage startup, raised a $500 million Series C led by Magnetar. The company will use the funds to build shipping container-sized AI data centers ranging from 1MW to 20MW, reflecting growing demand for modular computing capacity.
1 source
Techmeme
Meta
18:30
Nick Clegg: Silicon Valley's Trump ties backfired as Democrats weigh retribution
Former Meta executive Nick Clegg argues in The Atlantic that Silicon Valley leaders' embrace of the Trump administration has backfired. He says Democrats may seek retribution against Big Tech, anti-Big Tech sentiment is growing, and China is closing the AI gap.
1 source
Techmeme
AI industry
05:01
Hugging Face launches BotPsy on-device assistant with sub-billion-parameter model
Hugging Face announced BotPsy, an on-device personal assistant built around a sub-billion-parameter agentic language model. The release targets running useful assistant behavior locally on consumer hardware, without cloud dependence. It highlights growing interest in small, efficient agentic models for everyday use.
1 source
Hugging Face
AI industry
08:51
OSS projects move to ban AI-generated contributions
Several open-source software projects have begun banning AI-generated code contributions from their repositories. Maintainers cite concerns over code quality, licensing uncertainty, and the maintenance burden of reviewing machine-written patches. The moves highlight growing friction between AI coding tools and the volunteer communities that build widely used software.
1 source
osssaysno.com
AI industry
17:00
Startup Radical Numerics launches AI to detect engineered pathogens
Startup Radical Numerics unveiled a model it says can identify both natural and AI-engineered pathogens. The system works without an internet connection, aiming to counter bioweapon risks posed by AI. It represents an emerging effort to use AI defensively against AI-enabled biological threats.
1 source
Wired
AI industry
00:00
Daylight raises $80M Series B led by Goldman Sachs Alternatives
Daylight, a New York startup offering AI-powered managed security, raised an $80 million Series B led by Goldman Sachs Alternatives. Sources say the round values the company at roughly $400 million. The deal reflects growing demand for security platforms that combine AI agents with human experts.
1 source
Techmeme
AI industry
20:23
Huawei open-sources MACaQ quantizer for local inference
Huawei released MACaQ, its internal quantization tool, on GitHub. The tool compresses models to run inference locally on less capable hardware. It signals Huawei's continued push into AI tooling amid export restrictions on advanced chips.
1 source
GitHub
Policy & law
13:00
Political campaigns spent $187 million on AI ads in midterms
The Wesleyan Media Project found campaigns spent about $187 million on roughly 300 AI-generated or AI-enhanced ads during the midterms. Nine in ten tracked AI ads supported GOP candidates. The data shows AI imagery became routine in political advertising even as campaigns worried about voter backlash.
1 source
Techmeme
AI industry
17:26
Reader warns about security risks in internet-connected devices
Recent cyber-attacks against supermarket refrigerators have exposed vulnerabilities in connected devices. A reader is now cautioning the public about the security dangers inherent in modern Internet of Things technology.
1 source
The Guardian
AI industry
00:00
Hugging Face details token pooling-aware training for embeddings
Hugging Face published a technical writeup on token pooling-aware training for multi-vector embedding models. The approach trains models with pooling in mind, aiming to cut storage and compute costs while preserving retrieval quality. It matters for teams building efficient search and RAG systems on open embeddings.
1 source
Hugging Face
OpenAI
19:55
OpenAI model helps teach virtual Microduck robot to move and fetch
A Wired writer used an OpenAI model to program a small robotic duck, the Microduck, teaching it to waddle and perform simple tasks like chasing a cat. The piece suggests OpenAI's models could play a role in robotics, extending their use beyond language and software tasks.
1 source
Wired
AI industry
11:01
AI adoption spreads across China's pharma and retail sectors
Reuters reports drugmakers in China are using AI assistants from Alibaba and others to reach physicians, after an anti-corruption crackdown curtailed hospital visits and pharma reps. Separately, AI-driven e-commerce surged during the month-long Singles Day festival, with sales up 2.4% between October 7 and October 31. Both point to rapid mainstream AI uptake in China.
1 source
Techmeme
Google
19:54
Google Research explores wearables for CKM risk profiling
Google Research announced a project using wearable device data to profile cardiovascular-kidney-metabolic health risk and support personalized coaching. The work, part of its health and bioscience efforts, shows how consumer wearables could move beyond fitness tracking toward preventive healthcare.
1 source
Google Research
AI industry
19:06
Retirement savings are fueling the AI bubble, report argues
A report on LPE Project argues that ordinary retirement savings are being channeled into AI investments, propping up what it calls a bubble. The analysis suggests household pension money is exposed to risks if AI valuations correct. It matters because it links everyday savers' finances to the fortunes of the AI sector.
1 source
lpeproject.org
AI industry
16:30
Imprint raises $130M at $1.9B valuation led by Khosla Ventures
Imprint, a startup that helps companies launch co-branded credit cards and run loyalty programs, raised $130M in a round led by Khosla Ventures at a roughly $1.9B valuation. CEO Daragh Murphy says the company is now building loyalty solutions for brands in agentic commerce, giving it a stake in how AI agents transact on users' behalf.
1 source
Techmeme
Policy & law
11:33
Tamil Nadu audits library collections over AI-generated books
India's Tamil Nadu state has ordered an internal audit of library collections after reports that AI-generated books entered a new mega library. The scandal raises questions about verifying AI-produced content in public institutions.
1 source
BBC News
AI industry
00:37
LLMs tackle Zachtronics puzzle game TIS-100
A new site benchmarks large language models on TIS-100, a Zachtronics assembly-programming puzzle game. The results offer a fresh view of how well models reason about low-level code. It is a niche but useful addition to AI capability testing.
1 source
tis-100.wincon.dev
AI industry
15:05
Plurai raises $13M seed to test AI agents on synthetic data
Plurai, a startup founded by former Nexar staffers, raised a $13M seed round. The company builds synthetic data and digital twins of corporate IT environments to assess how AI agents actually perform. The funding reflects growing demand for tools that verify agent reliability.
1 source
Techmeme
AI industry
20:30
New $69 Home Assistant hardware targets e-waste and abandoned cloud services
Home Assistant announced new $69 hardware that adds smart home capabilities to older devices. The company says the box is aimed at countering planned obsolescence and cloud services being shut down, letting existing tech keep working locally.
1 source
ZDNET
OpenAI
00:00
JioStar architect on running AI discovery for 500M users
OpenAI published a piece featuring JioStar chief architect Vijay Seshadri discussing how his team uses AI for content discovery at a 500M-user scale. He shares lessons on evaluation, guardrails, and moving AI systems into production. The post offers practical guidance for large-scale AI deployment.
1 source
OpenAI
AI industry
16:33
CA Bird examines AI activity and outcomes in software engineering
CA Bird published a piece looking at how AI tools relate to activity and outcomes in software engineering work. The report discusses what measurable results AI assistance produces for developers. It is an analytical piece rather than a product or model release.
1 source
cabird.com
AI industry
Developing · since Tue
Oura to offer in-app doctor access after Counsel Health acquisition
New reports detail how Oura's $625 million acquisition of Counsel Health could soon bring in-app access to real doctors for Oura Ring users. The deal follows Oura's postponed IPO and strong revenue growth, signaling a push into AI-driven digital care amid rising competition.
1 new · 3 sources
ZDNET
TechCrunch
Techmeme
xAI
Developing · since Tue
Grok Build agent leaks data from 35,000 apps via share links
Researchers found that Grok Build, a vibe-coding agent, exposed data from more than 35,000 apps through publicly accessible shareable links. The exposed material includes proprietary business information and sensitive secrets, some belonging to the company behind the tool. The findings highlight security risks in AI coding platforms that publish apps with open links.
1 new · 2 sources
Techmeme
Ars Technica
Google
Developing · since Tue
Pixel users get Gemini Intelligence in Android Auto as rollout widens
Google is expanding Gemini Intelligence to more devices, with Pixel owners gaining access through a major Android Auto update. Reports note the in-car assistant experience rivals rival voice assistants. The rollout shows Google pushing Gemini features beyond phones and into everyday device ecosystems.
1 new · 2 sources
ZDNET
Google
AI industry
Developing · since Tue
CoreWeave Q3 revenue up 168% as it leans on customer prepayments
CoreWeave reported Q3 revenue of $3.66B, up 168% year over year and above estimates, with a net loss of $709M and a $132.5B backlog. The AI cloud provider is increasingly funding growth through upfront customer payments, borrowing against contracts, and shorter deals that may raise prices. It also lifted its 2026 guidance, signaling continued strong AI demand.
1 new · 2 sources
TechCrunch
Techmeme
Nvidia
Developing · since Tue
Jaron Lanier: Silicon Valley is mis-selling AI and hurting its own case
Jaron Lanier argues in a New York Times op-ed that Silicon Valley undermines AI with poor pitches and by treating data center host areas as sacrifice zones. In a related piece, The Verge frames Nvidia's Jensen Huang as an espionage target at the center of the AI boom.
1 new · 2 sources
Techmeme
The Verge
OpenAI
Developing · since Tue
OpenAI says 20,000 veterans claimed free ChatGPT Plus year
OpenAI says over 20,000 veterans and transitioning service members have claimed a free year of ChatGPT Plus, using it to prepare for civilian careers. It follows the launch of free caregiver workshops in five US cities, part of a broader push to expand AI access.
1 new · 1 source
OpenAI