The Wire
Daily AI news, learning & projects
News
Timeline
Models
Learning
GitHub
?
Latest
Daily
Social
Videos
‹ Sunday's edition
·
Daily edition
·
Tuesday ›
Monday 9 November
All
OpenAI
Anthropic
Google
Other
Anthropic
18:56
Anthropic launches Claude Fable 5.2 and Mythos 5.2, cuts prices 20%
Anthropic has released two new flagship models, Claude Fable 5.2 and Mythos 5.2, aimed at demanding technical work. Pricing drops 20% from the 5.1 generation to $8 per million input and $40 per million output tokens, with an optional faster mode costing double. Anthropic says the models improved coding, reasoning and vision, and topped Artificial Analysis' leaderboard with a score of 65.
5 sources
artificialanalysis.ai
Techmeme
TechCrunch
The Verge
Anthropic
Nvidia
Developing · since Thu
Trump says FBI probing Jensen Huang affair as Nvidia scandal widens
NEW
Trump said the FBI is looking into the affair, that Huang told him he had no idea the woman was a spy, and Huang pulled out of an ElevenLabs summit appearance.
President Trump said the FBI is investigating Nvidia CEO Jensen Huang's affair with Megaspeed executive Alice Huang, an alleged Chinese spy, and that Huang told him he was unaware of her ties. Nvidia shares slid over 4% amid board scrutiny, lawsuits and governance concerns.
7 new · 9 sources
The Guardian
Wired
The Verge
Techmeme
TechCrunch
BBC News
+3 more
AI industry
15:15
Medium launches TK, an AI-assisted writing and notes app
Medium launched TK, an app combining AI-assisted writing, note-taking and social feedback, its first major product since 2017. Users can invoke an AI helper in the editor with an @TK prompt for lookups, feedback and style-guide matching. The launch wades into publishing's debate over AI's role in writing, with disclosure tools for AI provenance.
3 sources
Simon Willison
TechCrunch
Techmeme
Nvidia
05:00
Nvidia Releases Dynamo Optimized Recipes for Korean Open AI Models
Nvidia has released deployment recipes leveraging the Dynamo framework for several major Korean open models. These optimizations enable high-performance, large-scale inference on NVIDIA B200 GPUs, accelerating the adoption of local AI across enterprise services.
2 sources
Techmeme
Nvidia
Meta
18:00
Meta releases Muse Ember, a lightweight model for agent execution
Meta has launched Muse Ember, a lightweight model designed for agent execution. Priced at 10 cents per million input tokens, it is significantly cheaper than frontier competitors such as Haiku and Luna. The low cost could make it attractive for developers running agents at scale.
2 sources
Techmeme
TechCrunch
Policy & law
15:50
Trump says DOJ threatened to jail a top AI executive
In a 60 Minutes interview, President Trump said the Justice Department told a leader of a major AI company to prepare for jail over allegedly bad activities with AI. He declined to name the person but claimed the executive changed course afterward. He also downplayed AI safety risks and said he would work with Democrats on guardrails.
2 sources
Techmeme
The Verge
AI industry
19:09
Apple launches $50 intelligence usage packs for extra AI access
Apple has begun selling one-time $50 intelligence usage packs for users who hit monthly caps on select Apple Intelligence features. The option, spotted in Logic Pro's Creator Studio apps, marks Apple's first paid tier for expanded AI usage. It signals Apple moving toward monetizing its AI features beyond the free allowance.
2 sources
Techmeme
The Verge
AI industry
15:38
Disney and Adobe partner on artist-led generative AI models
Disney and Adobe announced a multi-year partnership at Adobe's MAX conference to build proprietary generative AI models aimed at supporting Disney's production pipeline. The deal gives Disney early access to Adobe Firefly Foundry technology for use across theme parks, cruise lines, and ESPN graphics.
2 sources
Techmeme
The Verge
Google
08:00
Nvidia to invest in Korean startups after forum with government
...
2 sources
Nvidia
Wired
Google
18:24
Google Research unveils Language Explorer for world's languages
Google Research introduced Language Explorer, a natural language processing tool for exploring the world's languages. The project aims to broaden language coverage in NLP research. It could support linguistic analysis and technology development for underrepresented languages.
2 sources
Google Research
GitHub
AI industry
04:05
Cognition CISO Discusses How AI Is Redefining Cybersecurity
Cognition CISO Alex Stamos discussed how AI is impacting cybersecurity. He covered challenges like AI-powered red teaming and vulnerabilities found in older codebases. This reflects a critical shift in the modern security landscape.
1 source
Techmeme
Nvidia
16:20
Nvidia intervention shifted Megaspeed chip smuggling probe to DOJ
A Commerce Department investigation into Singapore firm Megaspeed, suspected of smuggling Nvidia AI chips to China, was moved to the Justice Department after Nvidia intervened, per reports. Bloomberg reports Nvidia's 2025 filings may have undercounted chips sold to Megaspeed via partners by roughly 50,000.
1 source
Techmeme
Google
12:50
Google unveils TPU 8tx and TPU 8i Max chips launching 2027
Google announced two specialized TPU variants: the TPU 8tx, designed to pair with AMD CPUs, and the TPU 8i Max, a dual-chip package aimed at long-context workloads. Both are slated for 2027 as Google broadens its AI chip lineup to capture growing demand.
1 source
Techmeme
Anthropic
Developing · since Sun
Anthropic buys $3.4B of gas turbines in bid to power AI growth
NEW
TechCrunch reports Anthropic's turbine purchase could leave it with buyer's remorse, adding a skeptical assessment of the $3.4B gas power bet.
Anthropic has bought commitments for 1.7GW of gas-fired generators from supplier OnePWR for $3.4B, reportedly the largest such purchase by an AI lab. The deal bypasses traditional utility suppliers as labs scramble to secure scarce power equipment amid surging AI energy demand.
1 new · 2 sources
TechCrunch
Techmeme
Policy & law
Developing · since Fri
Datacenter backlash seen as mostly an electricity fight
NEW
Reporting from Data Center Alley argues voter anger over datacenters is rooted mainly in electricity demand, framing the political backlash as an energy issue.
New analysis argues the backlash against AI datacenters, renewed by Democrats after the midterms, is largely about electricity demand. Elected officials, including Bernie Sanders and Gina Hinojosa, are pushing moratoriums that could slow the trillion-dollar buildout.
1 new · 3 sources
MIT Technology Review
Techmeme
The Guardian
AI industry
Developing · since Sun
Apple under pressure to speed up Siri's push toward agentic AI
NEW
ZDNET reports Apple's public satisfaction with the Siri AI launch hides internal pressure and tension over the upgrade's pace.
ZDNET reports internal pressure at Apple over the pace of Siri's AI upgrade, after Bloomberg detailed CEO John Ternus pushing Craig Federighi to accelerate the roadmap. Upgrades include fine-grained app control and third-party integrations, with a fuller AI agent possibly arriving next year.
1 new · 3 sources
ZDNET
The Verge
Techmeme
More from the day
AI industry
08:46
Ada and Dialpad Agree to Merge in All-Share Deal Valuing Ada At $430M
Toronto-based Ada and San Francisco-based Dialpad, both developers of AI customer service software, have announced a merger. This all-share deal values Ada at approximately $430 million to combine their technological resources.
1 source
Techmeme
Policy & law
15:40
EU draft law targets AI companion bots and addictive app features
A draft of the EU's Digital Fairness Act would require companies to let adults disable addictive features and target AI chatbots that simulate relationships. Per Bloomberg, the proposal extends EU regulation to social media platforms' engagement-driven design. It signals EU regulators' intent to directly govern AI companion products.
1 source
Techmeme
Policy & law
17:58
Palantir lobbied UK minister to extend AI tools to NHS GP records
WhatsApp messages show Palantir pressed health minister Zubir Ahmed MP to consider extending its AI platform into NHS GP records. Doctors fear such access to sensitive primary care data could undermine patient trust. The lobbying suggests the company is seeking to deepen its NHS footprint beyond hospital data.
1 source
The Guardian
Policy & law
17:05
Scott Wiener wins Pelosi's House seat after AI safety record
California state senator Scott Wiener, who authored legislation to curb risks from advanced AI, has narrowly won Nancy Pelosi's former congressional seat in San Francisco. He may now play a key role in shaping federal AI policy, giving the AI safety movement a potential advocate in Congress.
1 source
MIT Technology Review
AI industry
14:30
AI author platform Spines secures $29 million in equity financing
Spines, a platform utilizing AI agents for book editing and design, has successfully raised $29 million in equity. This funding provides the Herzliya-based startup with capital to continue expanding its comprehensive author services.
1 source
Techmeme
Google
14:00
Google criticized over pledge to pull Search from schools that block AI
Google said schools that block its new AI tools could lose access to Google Search, framing it as school choice over student AI use. Parents and teachers in the US are criticizing the policy, with some calling it extortion. The backlash tests how hard Google will push AI adoption in classrooms.
1 source
The Guardian
Meta
12:15
Meta expands Instagram AI age estimation globally for Teen Accounts
Meta announced a global rollout of Instagram's AI age estimation technology, applying Teen Account protections to users aged 13 to 17 even if they misrepresented their age. The move aims to extend safety features to millions more teens who claim to be adults on the platform.
1 source
Techmeme
Policy & law
05:15
Trump says he backs AI guardrails but not at cost of US competitiveness
President Trump addressed AI safety in comments published by CBS News, saying he does not mind guardrails but does not want the United States to fall far behind China, which he claimed has no such guardrails. The remarks signal a competitive framing of US AI policy.
1 source
Techmeme
Nvidia
05:10
Jensen Huang has placed family members in key Nvidia roles
Sources and filings reviewed by the Wall Street Journal show Nvidia CEO Jensen Huang has brought his children and other relatives into the company, making Nvidia unusually family-run for Silicon Valley. His children have been elevated to prominent positions. Nvidia's board is separately investigating reports that Huang was an espionage target.
1 source
Techmeme
Nvidia
14:00
Nvidia and partners show robots learning fine manipulation in simulation
At the Conference on Robot Learning, Nvidia and its partners presented research on using simulation to train robots for delicate tasks such as aligning gears, connecting cables and assembling small parts. Simulation was previously used mainly for locomotion. The work signals a shift toward robots mastering fine-grained manipulation before touching real hardware.
1 source
Nvidia
OpenAI
13:00
OpenAI unveils Control Room for overseeing AI workloads
OpenAI announced Control Room, a system that tracks model workloads across its research infrastructure. The tool can rapidly stop workloads when safety or security concerns arise, giving humans direct oversight of running AI systems.
1 source
OpenAI
AI industry
21:15
Fireworks says hackers accessed customer datasets and billing data
AI infrastructure provider Fireworks disclosed that hackers gained access to some customers' datasets, incomplete model weights, and billing information over the past month. Two customers told Forbes they had substantially moved their business off Fireworks' systems, underscoring security risks facing AI infrastructure providers.
1 source
Techmeme
OpenAI
19:00
ChatGPT Enterprise launches on AWS Marketplace
OpenAI has made ChatGPT Enterprise available through AWS Marketplace. Organizations can now purchase and bill for it using their existing AWS relationships, which is intended to simplify procurement for large customers.
1 source
OpenAI
AI industry
16:55
Samsung considers paid tiers for Galaxy AI features
Samsung is reportedly weighing usage limits and subscription tiers for Galaxy AI features that are currently free on its devices. Heavy users may need to pay for some AI capabilities. The move would mark a shift in how the company monetizes AI on Galaxy phones.
1 source
ZDNET
Meta
18:07
Instagram launches AI Effect Builder for custom Story filters
Instagram introduced Effect Builder, a tool that uses AI prompts to generate custom filters for Stories. It is available now to US residents over 18, following last week's macOS expansion of the Edits app. The feature shows Meta pushing generative AI deeper into consumer creative tools.
1 source
The Verge
AI industry
03:06
Asahi Linux switches to GPL after AI-assisted code theft
Asahi Linux, the project running Linux on Apple Silicon, says unreleased code was taken using AI and is relicensing its work from MIT to GPL. The change aims to require that derivative work stays open source. It highlights growing concerns about AI-assisted misuse of open-source projects.
1 source
itsfoss.com
AI industry
18:33
DeepX claims 100B-parameter inference under 5 watts
Chip startup DeepX says its hardware can run neural network inference for models up to 100 billion parameters at under 5 watts. If verified, the claim would enable large language models on edge devices without cloud dependence, a notable advance for low-power AI hardware.
1 source
eetimes.com
AI industry
12:34
Hugging Face proposes HOPE for keeping expert pairs together
Hugging Face published a post introducing HOPE, a method aimed at keeping paired experts together in mixture-of-experts models. The work addresses how expert pairs in MoE architectures can be preserved during training or transfer. It is a technical contribution relevant to researchers working on sparse model designs.
1 source
Hugging Face
AI industry
14:00
Automattic launches Spacefast vibe coding platform
Automattic debuted Spacefast, a platform for deploying vibe coding apps. The move positions the WordPress maker against rivals like Lovable as it seeks relevance amid a shifting internet landscape.
1 source
TechCrunch
Policy & law
14:30
Former antitrust chief Jonathan Kanter launches AI-focused firm
Former Justice Department antitrust chief Jonathan Kanter and ex-acting division chief Doha Mekki have launched a private-sector venture using AI to challenge illegal monopolies and Big Tech. The move extends Kanter's antitrust enforcement work after leaving government in 2024, signaling AI tools being applied to competition law.
1 source
The Verge
AI industry
11:00
Writers turn to paper trail tools amid AI provenance paranoia
Wired reports that authors are using paper trail tools to defend against accusations of undisclosed AI use in manuscripts. Publishers increasingly demand proof of how work was created, putting writers under pressure to document their process. The story highlights the industry's unresolved struggle over manuscript provenance.
1 source
Wired
AI industry
09:00
AI agents complicate how companies grade and pay managers
The Financial Times reports that firms deploying AI agents in place of staff face difficulty setting managers' grades and pay, which often depend on headcount. The issue surfaced in a panel discussion and highlights how AI adoption is straining traditional workplace structures.
1 source
Techmeme
Policy & law
22:19
Concerns raised over AI code comprehension after utilization mandate
A discussion surfaced regarding the inability to understand underlying AI code after a mandated integration. This highlights ongoing industry concerns about the complexity and transparency of deployed AI systems. The post points to significant knowledge gaps surrounding modern AI frameworks.
1 source
reddit.com
OpenAI
08:00
CodeRabbit uses OpenAI models and Codex to boost software output 20x
CodeRabbit said it uses OpenAI models to govern and build trust in AI-generated code, alongside Codex, its coding agent. The company reports shipping 20 times more software with the same engineering team. The announcement highlights how agentic coding tools are scaling real-world development workflows.
1 source
OpenAI
AI industry
05:30
Apple reportedly capped engineers' daily AI budgets at $200
Apple limited its engineers to spending no more than $200 per day on AI resources, according to a report. The cap suggests the company is controlling internal AI tooling costs as it manages budgets across its development teams. The policy was reported by Stork.
1 source
stork.ai
AI industry
17:46
Samsung launches Galaxy SmartTag 3 with next-gen Bluetooth tracking
Samsung released the Galaxy SmartTag 3, which supports emerging Bluetooth tracking technology. However, the tracker remains limited to Samsung's own ecosystem, excluding users of other platforms. The launch highlights Samsung's continued push in the Bluetooth tracker market.
1 source
ZDNET
AI industry
15:05
Hugging Face releases Nemotron-Personas-Finance-Japan synthetic dataset
Hugging Face published Nemotron-Personas-Finance-Japan, a synthetic dataset reflecting Japan's demographics and financial context. It gives researchers and model builders a Japan-focused resource for training and evaluating finance applications. The release expands availability of regionally grounded synthetic data.
1 source
Hugging Face
AI industry
11:50
Harwood, North Dakota voters back Applied Digital's $3B data center
Voters in Harwood, North Dakota rejected a ballot measure opposing Applied Digital's planned $3 billion data center, set to open in 2027. The result stands out against a broader nationwide backlash at the ballot box against large AI data center projects.
1 source
Techmeme
Policy & law
11:00
Guardian op-ed argues today's AI limits raise real catastrophic risks
A Guardian opinion column by Sunita Verma argues current AI systems will not reach superintelligence, but their technical limits still make them dangerous. It cites Geoffrey Hinton's estimate of at least 10% extinction risk and urges regulation over Silicon Valley focus on superintelligence scenarios.
1 source
The Guardian
Policy & law
11:00
Opinion: anti-AI populism may shape US politics after midterms
A Guardian opinion column argues the Democratic midterm victory signals rising anti-AI populism in American politics, since voters rebuked Donald Trump, the AI industry's most powerful champion. The writer urges Democrats to publicly call AI leaders like Sam Altman, Dario Amodei and Elon Musk to account for their ambitions. It is commentary, not news of a policy change.
1 source
The Guardian
AI industry
18:05
Show HN: STE80 tool rewrites AI text toward Simplified Technical English
A developer launched STE80 Humanizer on GitHub, a tool that rewrites AI-generated text by moving it 80% of the way toward the ASD-STE100 Simplified Technical English standard. The approach aims to strip out telltale AI phrasing and produce plainer, more direct prose. It matters as a fresh take on the growing demand for de-AI-ing machine-written text.
1 source
GitHub
OpenAI
08:00
Kuken Industries uses OpenAI API to search decades of drawings
Kuken Industries is using the OpenAI API to index 70 years of cooling tower drawings. Engineers can now locate the drawing needed for a repair in about 4.5 minutes, instead of nearly an hour. The case shows AI search tools cutting routine lookup time in industrial settings.
1 source
OpenAI
AI industry
19:40
Iran attributed to hack hitting French supermarket freezers
Iranian hackers disrupted defrosting systems at a French supermarket chain, an attack described as not particularly complex. The Guardian reports the incident has geopolitical context and is presumed linked to similar Iranian cyber-attacks across Europe, which are often intended to send political messages rather than cause lasting damage.
1 source
The Guardian
Policy & law
Developing · since Fri
US prosecutors say AI voice scam losses top $129m in 18 months
Federal prosecutors in New York say businesses have lost over $129m to scam calls using AI-generated voices impersonating executives and lawyers since July 2024. Three foreign nationals were indicted for allegedly convincing employees at major law firms to transfer millions. The case highlights the growing threat of AI-enabled corporate fraud.
1 new · 2 sources
Techmeme
The Guardian
Anthropic
Developing · since Sat
Claude Code settings files expose thousands of secrets on GitHub
Thousands of secrets including passwords and API keys are exposed in public GitHub files after developers approved Claude Code prompts with the 'Don't ask again' option, which lets the coding agent write credentials into settings files that later get pushed publicly. Earlier reporting documented over 1,100 affected files.
1 new · 2 sources
Ars Technica
apiiro.com
AI industry
Developing · since Fri
Tencent AI chief says AI bio risks strengthen case for closed models
Tencent's chief AI scientist Yao Shunyu warned that AI systems will soon be capable of mastering biology, making harm possible, and argued Anthropic's case for closed models is strong. This follows reports of AI-driven anxiety among white-collar workers, including an Anthropic researcher's high-profile resignation.
1 new · 2 sources
Techmeme
The Guardian