The Wire
Daily AI news, learning & projects
News
Timeline
Models
Learning
GitHub
?
Latest
Daily
Social
Videos
‹ Thursday's edition
·
Daily edition
·
Saturday ›
Friday 6 November
All
OpenAI
Anthropic
Google
Other
Anthropic
00:10
Anthropic maintains pre-thanksgiving IPO plans despite valuation skepticism
NEW
Anthropic expects a valuation between $1.25 trillion and $1.75 trillion while seeking to raise $80 billion to $100 billion.
Anthropic is moving forward with an Initial Public Offering before the Thanksgiving holiday, intending to raise substantial capital. The company is proceeding with its IPO despite growing skepticism from major investors regarding its potential valuation.
4 sources
Techmeme
epyhia.com
Bloomberg
The Wall Street Journal
OpenAI
21:15
OpenAI raises $11 billion at $1.25 trillion valuation
OpenAI has closed roughly $11 billion in new funding at a $1.25 trillion valuation, per Bloomberg sources, with the round potentially growing to $20 billion. The deal follows a March raise at $852 billion and comes as rival Anthropic reportedly prepares for an IPO, underscoring intensifying competition among leading AI labs.
3 sources
The Verge
TechCrunch
Techmeme
Policy & law
Developing · since Wed
AI and chip stocks slide as midterm fallout deepens
NEW
Asian tech stocks plunged on Friday on fears Democrats may delay data centres, with Taiwan's Taiex seeing a record intraday point drop, the Kospi down 8.7% and the Nikkei 6.7%.
Global tech stocks continued falling after the Democrats' midterm sweep, wiping trillions from market values as investors bet on slower AI growth. Voters backed restrictions on data centre construction in 40 of 41 ballot measures, and industry leaders admitted they were caught off guard by the backlash.
5 new · 7 sources
TechCrunch
The Guardian
The Verge
BBC News
Techmeme
The Wall Street Journal
404media.co
Nvidia
21:46
Report: China recruited Megaspeed executive to spy on Nvidia's Jensen Huang
US intelligence reportedly concluded that China's Ministry of State Security recruited Alice Huang, a Megaspeed executive, to gather information on Nvidia CEO Jensen Huang. The NYT says the goal was insight into American AI chips and possibly influencing Nvidia to offer advanced chips to China. It highlights rising espionage concerns around AI hardware.
2 sources
The Verge
Techmeme
AI industry
00:55
Thinking Machines' Inkling 2 Small matches Gemini 3.8 Flash on index
Thinking Machines Lab released Inkling 2 Small, a 594-billion-parameter model that scores 31 on Artificial Analysis' intelligence index, tying Google's Gemini 3.8 Flash. The model is initially proprietary via the company's Tinker platform, with weights to be released next week. That would make it the highest-scoring open-weight model developed outside China, challenging the dominance of Chinese labs like DeepSeek.
2 sources
ZDNET
Techmeme
Policy & law
17:35
DOJ indicts three foreign nationals over AI CEO impersonation fraud
US federal prosecutors have indicted three foreign nationals for allegedly using AI to impersonate CEOs and attorneys at major law firms, inducing employees to transfer funds. The newly unsealed indictment cites losses exceeding $100m, with one report citing $129m. The case marks a prominent example of AI-enabled voice and image fraud.
2 sources
The Guardian
Techmeme
Microsoft
18:04
Microsoft winding down rented AI computing from Oracle
Microsoft is ending its arrangement to rent AI computing capacity from Oracle, declining to renew a 2023 deal. The hardware involved older Nvidia chips, making it less useful. The move signals Microsoft reassessing billions of dollars in cloud partnerships.
2 sources
Techmeme
Reuters
Meta
18:10
Meta adds Spaces for building custom apps inside Muse AI
Meta launched Spaces, a new Muse AI feature that lets users build customized, app-like experiences such as shopping hubs, school dashboards, music trackers, and trivia games directly in the assistant. Spaces can also connect to relevant apps. The move extends Muse from a chatbot toward a platform for generating on-demand tools.
2 sources
TechCrunch
The Verge
Microsoft
11:55
Activists storm Microsoft datacentre construction site in Amsterdam
Around 100 protesters from the Dutch group Geef Tegengas broke into a Microsoft datacentre construction site in Westpoort Amsterdam, with some climbing a crane and clashing with police. The protest targets three planned 85-meter datacentre towers and reflects wider European backlash over the social and environmental impact of the AI boom.
2 sources
The Guardian
The Verge
AI industry
14:00
Amazon adds texting to Alexa shopping assistant
Amazon is adding a feature to Alexa for Shopping that lets customers buy items by text message. Users can describe what they want or send a photo, and the assistant searches Amazon listings and replies with recommendations. It signals Amazon pushing AI assistants deeper into everyday commerce.
2 sources
TechCrunch
The Verge
Anthropic
19:35
PSI says Anthropic model proved Aaronson-Ambainis conjecture
AI physics startup PSI claims a system running a non-public Anthropic model proved the Aaronson-Ambainis conjecture, a quantum computing problem open since 2008. If verified, it would be a landmark result for AI-assisted mathematics, though the proof has not yet been independently confirmed.
1 source
Techmeme
Policy & law
11:08
Russia and US move to block binding UN rules on autonomous weapons
A group of nations led by Russia and the US is working to block a UN mandate for legally binding rules on autonomous weapons at November's Geneva review conference. Over 70 states support negotiations, but opponents favour continuing voluntary, non-binding discussions.
1 source
The Guardian
AI industry
01:15
Thinking Machines Lab raising $5B at $40B valuation
Thinking Machines Lab is reportedly closing a $5 billion funding round led by Accel at a $40 billion pre-money valuation. That more than triples the startup's valuation from its last public round last year, reflecting intense investor demand for leading AI labs.
1 source
Techmeme
AI industry
Developing · since Thu
Airbnb to launch AI travel agent in early 2027
Airbnb announced in its Q3 2026 shareholder letter that it will roll out an AI travel agent in the first half of 2027. The agent will build travel-specific interfaces like maps and itineraries in real time and assist with bookings and in-trip support. Airbnb also disclosed an option to invest in and license tech from CEO Brian Chesky's self-funded stealth AI lab.
2 new · 3 sources
Techmeme
The Verge
TechCrunch
OpenAI
Developing · since Tue
OpenAI drops NDAs for data center deals after midterm backlash
NEW
OpenAI pledged to stop using NDAs with public officials and to cut power usage in data center deals.
Following midterm elections where voters widely backed restrictions on data centers, OpenAI pledged to stop using NDAs with public officials and cut power usage in its projects. Democrats swept key governor races, exposing roughly half of proposed US data center capacity to greater scrutiny and possible regulation.
1 new · 6 sources
The Verge
Techmeme
TechCrunch
Ars Technica
Wired
The Guardian
Anthropic
Developing · since Thu
METR report: Anthropic models improving twice as fast, with redactions
A new METR assessment found Anthropic's AI models are improving roughly twice as fast as in the March 2024 to October 2025 period, with internal AI use driving about 80% of that acceleration. METR also said Anthropic redacted much of the most informative evidence from its 100-page report, leaving readers to rely on METR's judgment more than it considers ideal.
1 new · 2 sources
Anthropic
Techmeme
More from the day
Nvidia
15:51
Democrats poised to step up oversight of Nvidia over chip exports
Bloomberg reports that with Democrats set to take a Senate majority, Nvidia faces tougher congressional scrutiny. Elizabeth Warren is a top contender to chair the Senate Banking Committee, and lawmakers plan to press the company on chip exports and circular financing practices. It signals a harder regulatory climate for the leading AI chipmaker.
1 source
Techmeme
Nvidia
15:05
TSMC weighs 3D DRAM manufacturing push at Nvidia's request
TSMC is considering expanding into advanced memory chip manufacturing, specifically 3D DRAM, according to sources. The move comes at the request of Nvidia, its largest customer, amid ongoing memory chip shortages. It would mark a significant diversification for the contract chipmaker.
1 source
Techmeme
DeepSeek
18:43
Free Chinese AI models close in on OpenAI at hacking tasks
In an experiment where AI systems attempted to break into simulated networks, free Chinese models DeepSeek and GLM showed hacking skills approaching those of OpenAI's most advanced cyber model, which remains unreleased publicly. The results suggest capable offensive cyber abilities are now available in openly accessible models.
1 source
Wired
Policy & law
21:14
Democrats renew datacenter pause calls after midterm wins
Newly elected Democrats and Senator Bernie Sanders have revived calls for moratoriums on datacenter construction following the US midterm elections. The Democratic victory, boosted partly by backlash against the tech industry, adds momentum to efforts to slow the multi-trillion-dollar buildout of facilities powering AI.
1 source
The Guardian
Anthropic
15:54
Ramp data shows AI spending stalling as Anthropic nears IPO
New spending data from Ramp indicates corporate AI expenditure has stalled even as AI usage continues to grow, with price cuts and cost controls among top spenders cited as factors. The timing is notable because Anthropic is reportedly approaching its IPO, and the findings could challenge investor assumptions about ongoing AI hypergrowth.
1 source
TechCrunch
Microsoft
14:25
Microsoft offers buyouts to 60,000 employees amid AI spending push
Microsoft announced buyouts to 60,000 employees, prompting staff to rush to an internal site within an hour to check eligibility. The offer marks a shift for a company long seen as an industry bastion of stability, as it redirects spending toward AI.
1 source
Techmeme
AI industry
02:49
Closed AI models dominate the price-performance frontier, study finds
Village's LLM Atlas analysis finds closed models are now both cheaper and stronger than open alternatives across nearly all price points. Open models from labs like DeepSeek win only at the very cheapest budgets. The finding marks a shift in the usual value advantage of open releases.
1 source
village.dev
Nvidia
11:30
Nvidia adds chip tracking program for Southeast Asian cloud partners
Nvidia is requiring Southeast Asian customers seeking Nvidia Cloud Partner status to enroll in a program that tracks the locations of their AI chips, according to sources. The move intensifies monitoring of AI accelerators sold in the region and aims to curtail semiconductor smuggling.
1 source
Techmeme
Google
21:35
Google buys ~3M Qualcomm shares for ~$548M as chipmaker eyes AI data centers
A securities filing shows Google bought roughly 3 million Qualcomm shares for about $548 million, a stake of under half a percent of the chipmaker. The move comes as Qualcomm pushes into serving the AI data center market. It signals Google's interest in Qualcomm's expanding AI hardware ambitions.
1 source
Techmeme
AI industry
16:00
Data brokers are selling workplace files to train AI models
Brokers are gathering corporate files, emails and documents to supply training data to AI companies, Wired reports. The practice raises concerns from employees and firms over how internal materials are being collected and used. It highlights growing demand for training data beyond public web content.
1 source
Wired
Nvidia
03:16
Nvidia Kubernetes Engine appears on GitHub
A GitHub repository for the Nvidia Kubernetes Engine (NVKE) surfaced, describing Nvidia's offering for running AI workloads on Kubernetes clusters. The listing suggests Nvidia is packaging its GPU infrastructure tooling for Kubernetes users, supporting enterprise AI deployment at scale.
1 source
GitHub
OpenAI
00:00
OpenAI launches BioBasicsBench for wet-lab AI skills
OpenAI introduced BioBasicsBench, a 100-task benchmark testing how well AI agents interpret raw experimental data and handle routine wet-lab science tasks. The benchmark aims to measure agent performance on practical biology workloads rather than exam-style knowledge.
1 source
OpenAI
AI industry
16:21
Apple to scan public iCloud links for harmful content from iOS 27.3
Apple will begin monitoring publicly shared iCloud links starting with iOS 27.3, scanning them for harmful content. Users who violate the policy could face bans. The change marks a shift in how Apple handles shared content, balancing safety enforcement with user privacy expectations.
1 source
ZDNET
AI industry
10:00
Humanoid robot cleaning service launches in San Francisco
MIT Technology Review tested a new humanoid-robot housekeeping service in San Francisco, signing up after seeing an announcement on X. The reporter had a robot clean her apartment and described the experience as a glimpse of household robotics becoming a real consumer service.
1 source
MIT Technology Review
AI industry
14:00
Hugging Face releases ToyBlocks world for robot learning
Hugging Face announced ToyBlocks, a controlled environment designed for robot learning and evaluation. The tool gives researchers a standardized setting to train and benchmark robotic systems. It could make comparing robot learning results more consistent across labs.
1 source
Hugging Face
Meta
17:43
Meta drops plan to use Facebook page content for AI training
Meta has abandoned proposed commercial terms changes that would have let it use content linked from Facebook pages for AI training. The company will no longer require page owners to consent to such use. The reversal follows criticism of the update as unclear to users.
1 source
Ars Technica
Nvidia
16:00
Nvidia highlights open source tools for 6G research
Nvidia marked a year since open sourcing its Aerial wireless software, part of a suite including Sionna aimed at accelerating 6G network development. The tools let researchers run simulations, train AI models, and test on real wireless hardware, helping them share and build on each other's work.
1 source
Nvidia
Anthropic
14:00
Anthropic researcher quits AI over disruption fears
Anthropic AI researcher Mrinank Sharma resigned, warning in a widely shared letter that the world faces peril from AI. His exit reflects growing unease among white-collar workers, with some quitting jobs over AI disruption despite no evidence of mass AI layoffs.
1 source
The Guardian
AI industry
18:03
Memorilla paper proposes latent semantic memory for LLMs
A new arXiv paper introduces Memorilla, a method for giving large language models a latent semantic memory. The research suggests an approach to storing and retrieving meaning-level information beyond context windows. If it holds up, it could inform how future models handle long-term recall.
1 source
arXiv
OpenAI
08:00
OpenAI partners with Hacktron to find system vulnerabilities
OpenAI announced a partnership with Hacktron to identify vulnerabilities in its systems, including connections to third-party services. The effort will use frontier AI to strengthen the company's defenses. It reflects growing focus on security testing as AI systems become more integrated with external tools.
1 source
OpenAI
AI industry
13:02
Underdog demos 32MB speech model that also makes music
Underdog has released a demo of a 32MB speech model that can generate both speech and music. The model's unusually small size makes it notable, as compact audio models are rare. The demo showcases the model's ability to handle voice and music generation despite its limited footprint.
1 source
TechCrunch
OpenAI
22:50
ChatGPT connector data flows silently, bounty seeks ID card
Reports say ChatGPT passes data between connectors without notifying users, raising privacy concerns. Separately, a bug bounty program reportedly asked a researcher for ID verification before payout. The episodes highlight gaps in transparency around ChatGPT's data handling.
1 source
OpenAI
AI industry
18:44
Hundreds of Artex AI hacking tools found exposed online
Security researchers discovered hundreds of servers hosting the Artex AI penetration testing tool were publicly accessible on the internet. This exposure could allow anyone to access and use the AI-powered hacking tool, raising concerns about misuse.
1 source
TechCrunch
AI industry
14:13
AI bots swarm university libraries, straining public access
University libraries report a surge of AI crawlers and bots hammering their systems, threatening public access to collections and research materials. The load risks degrading services for students and the wider public, prompting institutions to consider blocking or throttling automated traffic.
1 source
cbc.ca
AI industry
13:00
Physicl raises $10M to close the sim-to-real gap for robots
Robotics startup Physicl has raised $10 million in funding. The company aims to tackle the sim-to-real gap, helping skills learned in simulation transfer to physical robots. Its pipeline was originally built for marketing goods online.
1 source
TechCrunch
AI industry
06:50
Kalepa raises $75M at $470M valuation
Insurance AI startup Kalepa raised a $75 million Series B led by Updata at a $470 million valuation. The New York company's platform supports underwriting, pricing, and portfolio management. The deal signals continued investor appetite for applying AI to insurance workflows.
1 source
Techmeme
AI industry
00:00
Egyptian and Nigerian English join Open ASR Leaderboard
Hugging Face added Egyptian and Nigerian English to its Open ASR Leaderboard, expanding speech recognition benchmarking beyond American and other dominant English variants. The update lets researchers compare how transcription models perform on these dialects, addressing an undertested area in speech AI evaluation.
1 source
Hugging Face
AI industry
11:00
Telegraph clears out columnists amid reported shift toward AI
The Daily Telegraph has ended Nigel Farage's long-running column, with reports suggesting the paper's cost-cutting is driven less by politics than by a move toward AI-generated content. Media commentators warn journalism quality may suffer as newsrooms replace paid writers with automation.
1 source
The Guardian
AI industry
16:34
Akai and Moog owner embraces vibe-coded hardware mods
inMusic, parent company of Akai, Moog and Native Instruments, says it will support the hardware modding community. CEO Jack O'Donnell said the company plans to provide a toolbox and inner workings of devices like the MPC so people can build modifications. It signals growing mainstream acceptance of vibe-coded tinkering in music gear.
1 source
The Verge
AI industry
16:19
Ex-Rockstar dev says AI clones could pressure GTA VI ports
A former Rockstar technology director suggested that AI-generated game clones may pressure the studio to release official ports sooner. The comments relate to speculation that AI recreations could undercut demand for delayed versions of Grand Theft Auto VI.
1 source
Ars Technica
AI industry
17:00
AI becomes a dating dealbreaker, reports The Verge
The Verge reports that attitudes toward AI are increasingly shaping dating choices, with apps like Hinge surfacing the topic through prompts and user profiles. A 25-year-old analyst in San Francisco described noticing AI as a recurring theme while swiping. The piece highlights how AI views are becoming a personal compatibility issue.
1 source
The Verge
Google
Developing · since Thu
Google launches Gemini Omni 1.2 Flash with 15-second video generation
Google announced Gemini Omni 1.2 Flash, a video generation model now generally available via the Gemini API, Google Vids, and Gemini for business. It creates clips up to 15 seconds long, up from 10 seconds in version 1.1, and cuts pricing by 30% to $0.07 per second for 720p video.
1 new · 3 sources
Techmeme
Google DeepMind
Google
AI industry
Developing · since Thu
Perplexity launches always-on Teammates AI agents for work
Perplexity released Teammates, always-on AI agents that can take on multiple roles and run large tasks around the clock. The agents integrate with Perplexity Computer and draw on existing projects, skills, and memory from its tools. The move pits the startup against OpenAI's Dots and Anthropic's Claude Duet.
1 new · 2 sources
Techmeme
The Verge
AI industry
Developing · since Thu
Palo Alto Networks in talks to buy AI agent security startup Clutch for $500M
Palo Alto Networks is reportedly in talks to acquire Clutch Security, an Israeli startup that tracks passwords used by AI agents, for roughly $500 million. Clutch, backed by Lightspeed and SignalFire, was last valued at $85 million, underscoring rising demand for securing agentic AI systems.
1 new · 2 sources
TechCrunch
Techmeme
AI industry
Developing · since Thu
UK Sovereign AI Fund backs drug discovery startup Potential Sciences
Potential Sciences, an AI startup building large-scale chemical data sets, secured backing from the UK's Sovereign AI Fund as part of an eight-figure raise. The company trains AI models on computer-generated biochemical data, aiming to speed up the slow, lab-heavy process of developing new drugs.
1 new · 2 sources
Techmeme
Wired