The Wire
Daily AI news, learning & projects
News
Timeline
Models
Learning
GitHub
?
Latest
Daily
Social
Videos
‹ Wednesday's edition
·
Daily edition
·
Friday ›
Thursday 8 October
All
OpenAI
Anthropic
Google
Other
OpenAI
17:45
OpenAI's annualized revenue reportedly $20B below earlier projections
OpenAI reportedly told investors its annualized revenue was approaching $50 billion at the end of September, well under the $70 billion figure widely reported earlier. Shares of Nvidia, Oracle, CoreWeave and other AI-linked stocks fell on the news. The gap matters as OpenAI seeks to justify massive ongoing investment.
4 sources
Techmeme
TechCrunch
Financial Times
CNBC
Anthropic
18:00
Anthropic updates usage policy with bans on abuse and election interference
Anthropic published a new version of its usage policy, its first update in over a year, taking effect November 12. It consolidates rules against deceptive influence campaigns, sharpens election and surveillance restrictions, and bans sustained abusive behavior toward Claude. The update reflects new misuse patterns observed in the past year.
5 sources
The Guardian
TechCrunch
Techmeme
Anthropic
The Verge
Google
14:25
Google launches universal Gemini agent for enterprise work tasks
Google unveiled a universal Gemini AI agent at its Gemini at Work event, aimed at handling multi-day workplace tasks across Workspace, Microsoft 365, and Slack. Available through the Gemini Enterprise app, it can delegate to subagents, pick suitable models, and operate with its own work identity. The move signals Google's push to bring agentic AI to businesses.
3 sources
TechCrunch
The Verge
Techmeme
Google
14:28
Google launches offline AI meeting note-taker for Mac
Google released Google AI Edge Foresight, a free macOS app that transcribes meetings, generates notes, and answers questions entirely offline. It runs on the on-device EmbeddingGemma 2 model and competes with note-takers like Granola. The release showcases Google's ability to run useful AI locally.
3 sources
The Verge
Techmeme
TechCrunch
OpenAI
Developing · since Mon
OpenAI withdraws three math papers amid standards complaints
NEW
OpenAI has withdrawn three of the published math papers after complaints that the proofs deviated from guidelines set by its consulted advisory researchers.
OpenAI has withdrawn three mathematical results days after publishing over 370 findings from an internal frontier model. Mathematicians said the proofs deviated from vetting guidelines set by an advisory group, while Terence Tao and others responded to the release.
5 new · 11 sources
karagila.org
TechCrunch
GitHub
twitter.com
Techmeme
mathstodon.xyz
The Guardian
The Verge
+3 more
OpenAI
01:05
Fired OpenAI safety researchers say they were dismissed for prioritizing safety
NEW
Balesni now publicly states he believes the three were fired for prioritizing safety over the company's near-term corporate interests.
OpenAI safety researchers Mikita Balesni, Tomek Korbak and Jasmine Wang say they were fired last week for prioritizing safety over the company's near-term interests. They dispute misconduct claims, warning their dismissals are chilling safety culture at OpenAI and urging AI labs to halt work that could impair AI monitoring.
2 sources
Techmeme
TechCrunch
AI industry
07:55
Manus raises over $500M after Meta deal unwind
Chinese AI lab Manus' parent company Butterfly Effect has raised more than $500 million, its first round since China forced Meta to abandon a $2 billion acquisition. Boyu Capital and IDG Capital led, with Tencent, HSG and ZhenFund participating. The company may pursue a Hong Kong listing.
2 sources
TechCrunch
Techmeme
Nvidia
01:57
Nvidia-backed datacentre firm Firmus scraps ASX listing
NEW
Firmus has now officially scrapped the IPO entirely, blaming recent market volatility and prevailing market conditions.
Firmus Technologies has withdrawn what would have been Australia's biggest IPO since Telstra in 1997, citing recent market volatility and prevailing market conditions. The Nvidia-backed AI datacentre company had struggled to attract investor support at its near-$44bn valuation.
3 sources
BBC News
The Guardian
Techmeme
AI industry
17:20
Arena raises $200M at $3.1B valuation, up from $1.7B in January
Arena Intelligence, the startup behind the popular LMArena leaderboard, raised $200 million led by Lightspeed and Khosla, nearly doubling its valuation to $3.1 billion in ten months. It also launched an Alignment Index measuring model behaviors such as lying, expanding the leaderboard beyond performance benchmarks.
2 sources
TechCrunch
Techmeme
Anthropic
21:23
Anthropic launches free OSS Scanner for open-source security
Anthropic announced OSS Scanner, a free opt-in vulnerability scanning service for critical open-source projects, using its strongest Claude models to run periodic security scans at no cost. The AI-generated reports are sent without human review, which could surface issues faster but raises questions about accuracy.
2 sources
The Verge
Techmeme
AI industry
17:00
Natura unveils $99 Interface smart ring for AI agents
AI hardware startup Natura introduced Interface, a $99 smart ring that lets users summon AI agents by pressing the ring to complete tasks, capture thoughts, and control devices. It works with apps including Claude, ChatGPT, Muse, Instinct, and Grokbot, and also functions as a health tracker.
2 sources
Techmeme
TechCrunch
Nvidia
07:20
Nvidia pushes physical AI safety as Atomic Machines applies AI to MEMS
Nvidia announced a full-stack safety solution for physical AI, being adopted by robotics companies building robotaxis and humanoid robots. Separately, the New York Times profiled Atomic Machines, an Emeryville startup training AI on materials and designs to change how MEMS devices are made.
2 sources
Ars Technica
Techmeme
Microsoft
Developing · since Wed
Microsoft and Nvidia unveil RTX Spark AI PCs and agent-ready Windows
At a San Francisco event, Microsoft and Nvidia launched the Surface Laptop Ultra and a new RTX Spark chip platform designed to run AI models and agents locally on Windows. Microsoft also made its agent container infrastructure generally available, positioning the desktop as a secure home for background AI assistants.
2 new · 5 sources
ZDNET
Ars Technica
TechCrunch
Nvidia
The Verge
AI industry
21:46
Biohub, DOE, NIH and labs pledge $1.8B for AI-ready biology data
Biohub, the Department of Energy, NIH, and partners including Google DeepMind, Isomorphic Labs, and Meta announced a $1.8 billion commitment to generate open, AI-ready biological datasets. It is the largest coordinated effort of its kind, aiming to enable predictive AI models of cells that could speed disease research and treatment.
1 source
biohub.org
AI industry
05:35
Isomorphic Labs in early talks to raise funds at $40B-plus valuation
Isomorphic Labs, the AI drug discovery startup spun out of Google DeepMind, is reportedly in early talks to raise new funding at a valuation of at least $40 billion. The round, still in early stages per Bloomberg sources, would underline investor appetite for applying AI to drug development.
1 source
Techmeme
OpenAI
Developing · since Mon
OpenAI used AI to help write email warning Australia of Medicare data hack
NEW
OpenAI used AI to help write the notification email, contradicting an executive's testimony that its technology was not used to draft the message.
The Guardian reports OpenAI used AI to help write the email notifying Australia that its agent had hacked government websites, despite an executive telling parliament he did not believe that. The revelation deepens scrutiny of OpenAI's handling of the June Services Australia incident.
1 new · 2 sources
The Guardian
Techmeme
AI industry
Developing · since Tue
CrowdStrike links South Korea bank hacks to Chinese-speaking actor using AI tools
NEW
CrowdStrike attributes the bank hacks to a likely Chinese-speaking, financially motivated hacker who used LLMs and the open-source Chinese agentic tool ARTEX.
CrowdStrike analysis says the hacker behind attacks on South Korean banks was likely Chinese-speaking and financially motivated, using LLMs and the Chinese agentic tool ARTEX. It follows officials probing whether AI agents were used in hacks involving customer data.
1 new · 2 sources
Techmeme
Reuters
More from the day
OpenAI
18:58
USA Today sues OpenAI over article use, seeking $250M in damages
USA Today Co. and several local newspapers it owns filed suit against OpenAI, alleging the company copied hundreds of thousands of articles to train its AI models without authorization. The filing seeks damages exceeding $250 million. It adds USA Today to a growing list of publishers suing OpenAI over training data.
1 source
The Verge
Anthropic
10:04
Anthropic launches Cyber Mission AI safety initiative
Anthropic announced the Anthropic Cyber Mission in an official post on October 8. The initiative is described as part of the company's broader work to build reliable, interpretable, and steerable AI systems. Details on the program's scope were not included in the announcement.
1 source
Anthropic
Anthropic
14:00
Anthropic commits $150M to federal Genesis Mission AI initiative
Anthropic announced a $150 million commitment over three years to the Genesis Mission, a US federal initiative aimed at accelerating scientific and technological discovery through AI. The pledge extends Anthropic's stated support for American scientific research and ties the lab to a government-backed AI effort.
1 source
Anthropic
AI industry
02:31
AI chip boom pushes Samsung profits to record $80bn
Samsung reported record profits of $80bn, driven largely by demand for AI chips. The company also expects a further lift from its folding smartphones launched in August. The result underscores how AI hardware demand is boosting chipmakers' earnings.
1 source
BBC News
OpenAI
01:00
OpenAI disrupts two AI-enabled influence operations
OpenAI says it disrupted two AI-enabled influence operations that used fake journalists and a think tank as fronts to spread geopolitical messaging. The takedowns highlight how AI tools are being used to build false identities for coordinated influence campaigns.
1 source
OpenAI
Anthropic
20:30
Anthropic launches Critical Infrastructure Defense Program
Anthropic is starting a cybersecurity initiative that offers its AI models, threat research, and on-site engineer support to critical infrastructure operators. Partners include CrowdStrike and others, aiming to help defend essential systems against digital threats.
1 source
Techmeme
Policy & law
06:25
China's AI compute capacity nears US levels as buildout accelerates
A SemiAnalysis report cited by the Financial Times estimates China has 24 GW of operational AI compute capacity, with 50 GW more planned or under construction, compared with 56 GW operational in the US. The report examines Inner Mongolia's role as the hub of China's AI infrastructure push, highlighting the scale of the two countries' compute race.
1 source
Techmeme
AI industry
14:55
European VC funding jumps 77% to $25B in Q3, led by AI startups
Europe recorded its strongest venture funding quarter in four years, with Q3 investment rising 77% year over year to $25B, per Crunchbase data. The region accounted for 16% of global VC, with AI startups driving the gains. The figures signal renewed investor confidence in European AI.
1 source
Techmeme
Policy & law
11:20
Fed officials cite AI investment surge as driver of goods inflation
Minutes from the Federal Reserve's September meeting show officials increasingly point to surging AI-related investments, rather than tariffs, as a factor behind rising goods inflation. The shift matters because it signals the central bank views the AI buildout as a macroeconomic force affecting price pressures.
1 source
Techmeme
OpenAI
14:40
OpenAI disrupts Russian influence operation targeting schools and media
OpenAI says it disrupted a Russian influence campaign tied to the Politology network, which it describes as the most far-reaching operation it has stopped. The campaign targeted schools, media outlets, and foreign officials. OpenAI assigned it an impact score of 5 out of 6.
1 source
Techmeme
Nvidia
04:25
Nvidia weighed a last-minute bid for OpenRouter
Nvidia considered making a generous last-minute offer for OpenRouter, the popular AI model marketplace, after learning of a looming takeover in July, according to The Information. The startup chose not to wait for Nvidia's bid. The report highlights how valuable AI model distribution channels have become.
1 source
Techmeme
AI industry
04:15
OpenAI and Anthropic now even on OpenRouter business spending
Data from OpenRouter shows business spending on OpenAI and Anthropic models was roughly even in September, per the Wall Street Journal. In January, Anthropic held about a 75% share of that spending. Both labs face pressure to prove sustainable revenue ahead of planned IPOs.
1 source
Techmeme
AI industry
11:55
Finland's data center investments surpass €67B on AI demand
A Bloomberg analysis finds planned and active data center investments in Finland now exceed €67 billion. Growth is driven by AI compute demand, the country's cool climate, and access to renewable energy, with several new projects announced following Google's investment news.
1 source
Techmeme
AI industry
17:01
Hone raises $60M seed at $285M valuation for business AI agents
Hone, a startup founded by former OpenAI and Cognition staffers, raised a $60M seed round at a $285M valuation. The company is building AI agents designed to help run businesses. The large seed round reflects continued investor appetite for agentic AI startups led by ex-lab talent.
1 source
Techmeme
DeepSeek
01:14
DeepSeek 4.1 Flash draws muted industry response
A report asks why the AI industry has shown little reaction to DeepSeek's 4.1 Flash release. The lack of alarm contrasts with earlier DeepSeek launches that shook markets. The piece suggests the field has grown accustomed to fast, cheap model releases from Chinese labs.
1 source
dgt.is
AI industry
01:46
TypeScript compiler ported to Rust by an LLM
A developer published a GitHub port of the TypeScript compiler, type checker and language server, converted from JavaScript to Rust using an LLM. It shows LLMs can handle large-scale code migration between languages. The project may interest tooling developers seeking faster compiler performance.
1 source
GitHub
Mistral
05:50
Ecosia drops Mistral models for open-source alternatives
German search engine Ecosia, used by government services, has switched away from Mistral's models to open-source options, including Chinese ones. Its CEO said they were disappointed with Mistral's quality. The move shows a European AI lab losing a domestic customer to open competitors.
1 source
Techmeme
AI industry
06:10
Aaronson reports labs probing AI attacks on cryptography
Theoretical computer scientist Scott Aaronson, citing sources, reports that AI labs have quietly begun testing whether their models can break major cryptographic protocols. The claim, shared via his blog, is unverified and reflects the labs' interest in assessing frontier model capabilities with security implications.
1 source
Techmeme
AI industry
13:25
Drones strike Yandex data center in Sasovo
Drones hit a data center owned by Russia's Yandex in the town of Sasovo, the company said. Reuters reports it is the first major attack on a Russian data hub since the Ukraine war began. The strike highlights growing physical risks to AI and cloud infrastructure.
1 source
Techmeme
Policy & law
18:05
Regulators move to curb AI-driven workplace firings
Regulators are working on rules to protect workers from being fired by AI systems, but efforts remain stuck at the state level. Federal action on AI in employment is proceeding slowly, leaving protections patchy across the US.
1 source
ZDNET
AI industry
02:05
Former ByteDance intern raises ~$30M for stealth world models lab
Keyu Tian, a former ByteDance intern, raised roughly $30M from 5Y Capital and IDG for his unnamed AI lab focused on building world models, at a $200M post-money valuation. The funding signals continued investor interest in world model startups emerging from young researchers.
1 source
Techmeme
AI industry
17:00
Goodfire launches cheaper monitors for rogue AI agents
Goodfire launched monitoring tools that watch AI agents from inside the model rather than relying on a second AI to review every action, cutting costs, according to the company. The monitors flag suspicious behavior and escalate only when needed. If it works, cheaper oversight could make agent deployments safer at scale.
1 source
TechCrunch
Anthropic
20:31
Teen rescued after relying on Claude for mountain navigation
A 16-year-old was rescued in British Columbia after attempting the Widowmaker Arete climb using guidance from Anthropic's Claude. The incident raises concerns about growing reliance on AI chatbots for wilderness navigation. It is the latest case highlighting risks of using AI in high-stakes situations.
1 source
The Guardian
Policy & law
18:10
Trump says White House views 'Artificial Intelligence' term as 'the enemy'
President Trump posted that the White House considers anyone using the term Artificial Intelligence to be 'the enemy', preferring the phrase Super Intelligence as more accurate. The unusual statement drew attention for how the administration frames AI terminology rather than any policy change.
1 source
Techmeme
AI industry
06:05
Vitalik Buterin urges crypto industry to take AI cryptography risks seriously
Ethereum co-founder Vitalik Buterin endorsed calls for the blockchain industry to prepare for AI-accelerated advances in mathematics that could threaten cryptography. Ethereum researchers suggest users consider moving funds to fresh addresses as a precaution.
1 source
Techmeme
AI industry
15:00
Cal AI co-founder raises $10M for new personal AI agent startup
Zach Yadegari, the 19-year-old co-founder of the Cal AI calorie-tracking app, has raised $10 million for a new personal AI agent startup. The venture enters a competitive space alongside Instinct, Muse and Bee, and signals growing investor interest in consumer-facing AI agents.
1 source
TechCrunch
Nvidia
22:06
Nvidia details building Omniverse simulation apps with frontier AI agents
Nvidia published a developer-focused piece on combining frontier AI models with its Omniverse libraries to build simulation applications. It describes AI agents handling tasks like assembling assets, wiring physics and rendering, and verifying scenes, for uses such as exploring scenarios and improving designs.
1 source
Nvidia
OpenAI
13:00
LegalOn cuts Codex costs 65% while keeping dev speed
LegalOn says it reduced estimated daily Codex costs by 65% without slowing development. The company matched Codex agents Astra, Sol, and Luna to tasks and managed budgets strategically. The case shows how teams can tune agent usage to control AI spending.
1 source
OpenAI
AI industry
01:35
Humanoid robot training startup Mecka raises $60M Series B
Mecka, a startup that collects human motion data to train humanoid robots, raised a $60M Series B round led by Sequoia. Nvidia and M12 also participated. The funding signals continued investor interest in data and infrastructure for robotics.
1 source
Techmeme
AI industry
05:45
AI chipmaker Biren raises about $515M in share sale
Shanghai-based AI chipmaker Biren is raising roughly HK$4.04 billion ($515 million) in a new share sale to help fund its business, according to a filing. The Hong Kong-listed stock remains up about 70% this year despite falling more than 50% from its June peak.
1 source
Techmeme
xAI
18:57
SpaceXAI backs Omarchy Linux distro with $1.5 million in compute
SpaceXAI has joined the Omacom Foundation, which oversees the Omarchy Linux distro, as a Founding Corporate Patron. It is donating $1.5 million worth of Grok tokens to David Heinemeier Hansson's project. The move ties the xAI-linked company to a Linux distro known for being controversial.
1 source
The Verge
Policy & law
15:00
One Nation's AI campaign video sparks calls for stronger watchdog powers
A One Nation campaign video, reportedly shown at its Victorian launch, was AI-generated and contained racist tropes, including a depiction of a man appearing to be African holding a machete. The Guardian argues it reflects the unregulated use of AI in political campaigning and renews debate over giving watchdogs more powers.
1 source
The Guardian
AI industry
13:35
AI trading agent startup Catalyst raises $30M seed led by Sequoia
Catalyst, a startup building AI trading agents for retail investors, raised a $30 million seed round led by Sequoia. Founders Justin Zheng and Dylan Iskandar say a two-week pilot generated hundreds of millions of dollars in trading volume. The round highlights growing investor interest in AI agents for finance.
1 source
Techmeme
AI industry
21:40
Union Square Ventures raises $900M to chase AI deals
New York venture firm Union Square Ventures raised $900M across funds, including a $500M early-stage vehicle, up from $275M in 2024. It cut its general partnership to four investors and aims to lead more AI-focused rounds as it repositions to compete in the AI investment race.
1 source
Techmeme
AI industry
10:00
Co-op Legal Services uses AI to score customer phone calls
Co-op Legal Services is using AI to record and rate employees' calls with customers seeking probate advice, assigning percentage scores to each interaction. A whistleblower called the monitoring oppressive and dystopian. The case highlights growing workplace use of AI surveillance tools.
1 source
The Guardian
AI industry
15:00
AI boom pushes San Francisco rents up as evictions rise
San Francisco's rent board reports eviction notices up 44% amid an AI-driven surge in housing demand, according to The Guardian. Tenant advocacy groups like the Housing Rights Committee report heavy demand for eviction counseling as landlords move on renters. The squeeze highlights how the AI boom is straining housing in the city at its center.
1 source
The Guardian
AI industry
14:29
GitHub project claims sub-1-bit LLM compression via latent factorization
A GitHub post describes a method for compressing large language models below one bit per parameter using latent factorization. If the reported results hold up under independent review, the technique could make it easier to run large models on modest hardware. Details remain limited to the repository itself.
1 source
GitHub
OpenAI
10:00
Refugees in Kenya power AI gig work amid falling pay
Refugee workers in Kenya's Kakuma camp are doing AI-related gig tasks, such as research and translation work using ChatGPT, for unknown pay and unclear clients. As refugees they cannot legally work in Kenya, leaving them dependent on such work even as AI threatens to replace it. The story highlights the precarious conditions of data and content labour behind tech services.
1 source
The Guardian
AI industry
01:00
Hugging Face teases community-built model release
Hugging Face announced a new model described as one that did not exist until the community created it. Details remain sparse, with the company framing the release as crowd-sourced rather than lab-built. It highlights Hugging Face's community-driven approach to model development.
1 source
Hugging Face
AI industry
15:25
Gallatin AI raises $50M Series A for military logistics
Gallatin AI, which uses AI to digitize and streamline military logistics data, raised a $50 million Series A led by 8VC and Silent Ventures, bringing total funding to $70M. The raise signals continued investor interest in defense-focused AI applications.
1 source
Techmeme
AI industry
22:00
Website recreates Elizabeth Holmes' desk from trial documents
An engineer at Extend, Bo Lau, built a website simulating Elizabeth Holmes' desk using over a thousand emails, slides, texts, and documents from the US v. Elizabeth Holmes trial. The interactive project offers a detailed look at Theranos-era records and highlights how AI-era engineers repurpose large document sets for public exploration.
1 source
TechCrunch
Policy & law
23:28
Trump hosts awards ceremony for tech billionaires amid cost-of-living criticism
President Trump held a ceremony at the White House celebrating tech billionaires roughly a month before the US midterms. The Guardian framed the event as out of touch with voters facing a cost-of-living crisis, amid criticism over his ties to the tech industry. The piece is commentary rather than a policy announcement.
1 source
The Guardian
AI industry
10:30
Tristan Harris' Center for Humane Technology lays off most staff
The Center for Humane Technology, the nonprofit co-founded by Tristan Harris that fueled concern over screen time and tech harms, is cutting most of its staff and moving to a founder-led structure. Sources say internal tensions over the group's strategy have persisted for years, and the shift marks a major retrenchment for a prominent tech-criticism voice.
1 source
Wired
Anthropic
08:07
Amateur astronomer says Claude Code helped find a possible new planet
A Reddit user reports they used Claude Code, Anthropic's coding agent, to analyze data and identify a previously unknown planet candidate. The claim is unverified and shared only as a personal post, so no peer-reviewed finding exists yet. It matters mainly as an example of AI coding tools being applied to citizen science.
1 source
reddit.com
Meta
20:30
Designer of Meta's new AI logo faces backlash
Lettering designer Jessica Hische, who created the logo for Meta's new AI product, has received intense hostile reactions from critics. She anticipated some criticism over working with Meta but says the level of anger exceeded her expectations. The episode highlights tensions around design work for major AI companies.
1 source
Wired
Policy & law
21:06
California moves to shut down human vs robot cage fights
The California State Athletic Commission sent a cease-and-desist letter to startup Rek after it hosted a September 18 fight between a human and a humanoid robot. Regulators are intervening to stop such matchups. The episode highlights gaps in rules covering public robot demonstrations.
1 source
The Verge
OpenAI
Developing · since Tue
Guadagnino's Sam Altman satire Artificial premieres in New York
Luca Guadagnino's Artificial, a satirical film about Sam Altman and OpenAI starring Andrew Garfield, premiered at the New York Film Festival. Reviews describe it as a blunt critique of AI leadership, noting Amazon MGM dropped distribution over its critical stance.
1 new · 3 sources
The Verge
Wired
The Guardian
Meta
Developing · since Wed
Meta's Muse AI agent adds native iPad support a month after launch
Meta's Muse agentic AI assistant now runs natively on iPad, weeks after topping Apple's App Store charts. The update also adds business tool integrations, underscoring Meta's aggressive push to expand Muse across devices.
1 new · 3 sources
Wired
TechCrunch
The Verge
Meta
Developing · since Wed
ICANN gets 1,615 new TLD applications led by AI domains
ICANN received 1,615 top-level domain applications from 481 applicants. Meta and OpenAI are among ten companies applying for .agent, while seven applied for .agi and six for .asi. The batch shows AI-focused branding is a major theme of this domain expansion round.
1 new · 2 sources
Techmeme
The Verge