The Wire
Daily AI news, learning & projects
News
Timeline
Models
Learning
GitHub
?
Latest
Daily
Social
Videos
‹ Friday's edition
·
Daily edition
·
Sunday ›
Saturday 7 November
All
OpenAI
Anthropic
Google
Other
Policy & law
14:30
Chinese firms approved for Nvidia H200 chips had empty offices
A New York Times investigation found the Commerce Department approved dozens of Chinese firms to receive Nvidia's H200 AI chips, including companies with empty offices and ties to sanctioned entities. One listed firm, New Matrix, had an unoccupied office, raising concerns that vetting of approved buyers is inadequate.
2 sources
The Verge
Techmeme
Microsoft
18:02
Microsoft AI Frontiers head John Langford departs
Microsoft Research veteran John Langford has left the company, and reports say his AI Frontiers group has been wound down. Langford led reinforcement learning research at Microsoft. The exit marks the end of a notable research unit within Microsoft's AI organization.
2 sources
hunch.net
businessexaminer.ca
Policy & law
Developing · since Wed
Tech stocks stabilise after post-midterm AI rout
Tech stocks partially recovered after trillions were wiped from market values following the Democrats' midterm sweep. Voters backed restrictions on data centres in 40 of 41 ballot measures, and newly elected governors in states holding roughly 40% of planned capacity have pledged to pause construction.
1 new · 8 sources
kadoa.com
TechCrunch
The Guardian
The Verge
BBC News
Techmeme
The Wall Street Journal
+1 more
OpenAI
Developing · since Fri
Impagliazzo calls OpenAI's disproof of hardness conjecture shocking
Russell Impagliazzo, co-proposer of the Exponential Time Hypothesis, called OpenAI's claimed disproof of it shocking, saying it could eventually speed chip design and break encryption. The claim follows PSI's report that a system running a non-public Anthropic model proved the Aaronson-Ambainis conjecture, a quantum computing question open since 2008.
1 new · 1 source
Techmeme
Anthropic
Developing · since Thu
METR report: internal AI use doubling Anthropic's development pace
NEW
The Verge adds that the findings follow an 11-week METR investigation, with Anthropic's gains showing little sign of slowing.
A METR report says Anthropic's models are improving about twice as fast as between March 2024 and October 2025, with internal AI use driving roughly 80% of the acceleration. Anthropic redacted much of the report's most informative evidence, limiting public scrutiny. Researchers find the self-accelerating development trend alarming.
1 new · 3 sources
The Verge
Anthropic
Techmeme
Nvidia
Developing · since Thu
Memo earlier this year claimed Chinese spy agency recruited exec to target Nvidia's Huang
NEW
A memo circulating earlier this year made the same espionage claim about Huang being recruited by the MSS to target Nvidia's CEO, predating the recent intelligence reporting.
A memo circulated earlier this year claimed China's Ministry of State Security recruited Megaspeed executive Alice Huang to spy on Nvidia CEO Jensen Huang, per a TechCrunch report. US intelligence reportedly made the same conclusion, and House Republican leaders have asked Huang for a briefing on the matter.
1 new · 3 sources
TechCrunch
The Verge
Techmeme
More from the day
Meta
01:50
Meta ships Muse Wonder-trained models with Active Memory and Focus
Meta has introduced specialized versions of its existing models trained with Muse Wonder, adding Active Memory and Active Focus features for all Muse and Meta AI users. Wonder itself remains reserved for harder subscriber tasks. The move brings distilled frontier capabilities to the broader user base.
1 source
Techmeme
Anthropic
20:45
Claude Code's 'Don't Ask Again' option tied to exposed secrets in 1,122 GitHub files
Researchers found over 1,100 public GitHub files containing credentials after users reportedly approved Claude Code's permission prompts with 'Don't Ask Again'. The option can let the coding agent write secrets like API keys and passwords into files that developers later push publicly. It highlights risks of broadly trusting agent permissions in development workflows.
1 source
apiiro.com
AI industry
05:45
Sand.ai raises $200M for video generation
Beijing video generation startup Sand.ai announced a $200M Series A+, bringing its funding to $300M over the past 10 months. The round underscores continued investor interest in Chinese AI video companies.
1 source
Techmeme
Google
01:40
Crusoe sells troubled Wyoming AI data center project to Google
Crusoff has sold its troubled Wyoming data center project to Google. The deal highlights the difficulties of building the massive facilities needed for AI development. Terms of the sale were not disclosed.
1 source
TechCrunch
Policy & law
16:35
Democratic wins may pace AI investment, Deutsche Bank says
Bloomberg reports that Democratic election wins could slow AI investment by curbing data-center expansion in states the party won. Analysts at Deutsche Bank say guardrails may pace spending over time and shield the economy from a downturn by ending shaky projects.
1 source
Techmeme
Policy & law
10:00
First US digital air traffic control towers go live in Florida and Alaska
The FAA has made the first digital air traffic control towers operational at US airports, starting with Bartow executive airport in Florida and sites in Alaska. The systems offer 360-degree views of airport operations as part of efforts to modernise ageing aviation infrastructure, though some concerns about the technology have been raised.
1 source
The Guardian
AI industry
10:00
Power and water strains stall south-east Asia's datacentre boom
Thailand's datacentre expansion, buoyed by roughly $27bn in approved projects from TikTok, Skyline and Bridge datacentres, has slowed sharply. Power and water constraints are pressuring a region that months earlier appeared set for rapid growth in AI infrastructure.
1 source
The Guardian
Policy & law
07:40
FT examines why AI pessimism runs stronger in the West than in Asia
A Financial Times column by Gillian Tett argues that economic conditions, public policy, and cultural differences may explain why Western publics are more pessimistic about AI than those in Asia and emerging economies. It notes voter hostility to Big Tech influenced the US midterms, with similar fears spreading across the West.
1 source
Techmeme
AI industry
11:40
Nasscom president warns AI may squeeze fresh Indian graduates
Nasscom President Rajesh Nambiar said AI automation of entry-level coding could make it harder for recent Indian engineering graduates to land tech jobs. He called for an immediate overhaul of computer science curricula to address the shift.
1 source
Techmeme
AI industry
11:00
Caregiving apps add AI, but caregiver strain persists
Apps and platforms that help family caregivers manage loved ones' needs are gaining AI features that make them more capable. However, the wider problem of caregiver well-being remains unaddressed, suggesting the tools alone are not enough to ease the burden.
1 source
Wired
AI industry
15:30
Ocient raises $100M Series C to grow agentic AI analytics
Chicago data analytics firm Ocient raised a $100 million Series C co-led by Boka Capital and OCA Ventures. The company says it will use the funds to expand its agentic AI analytics business, signaling continued investor interest in applying AI agents to enterprise data analysis.
1 source
Techmeme
AI industry
07:00
Two-year test puts AI heating savings claims under scrutiny
A Guardian writer spent two years testing smart thermostats and AI heating controls against manufacturer claims of roughly 25% savings, or £180 a year. The trial aimed to separate marketing promises from real-world results amid changing energy costs. The findings surprised the tester.
1 source
The Guardian
AI industry
14:00
Philadelphia woman blocks food delivery robots in protest
Semi-retired professor Carolyn Marvin, 78, is blocking food delivery robots in Philadelphia, citing her right to protest and gaining local attention. The story highlights growing friction over the spread of sidewalk delivery robots in city neighbourhoods.
1 source
The Guardian
Policy & law
Developing · since Fri
Democrats renew datacenter pause calls after midterms; SpaceX builds Texas supercomputer
Newly elected Democrats and Bernie Sanders are pressing for moratoriums on AI datacenter construction after midterm gains fueled partly by tech backlash. Meanwhile, public records suggest SpaceX is building a Starbase supercomputer in Texas, though governor-elect Gina Hinojosa may delay some datacenter plans.
1 new · 2 sources
Techmeme
The Guardian
AI industry
Developing · since Fri
Hundreds of AI hacking tool Artex copies found exposed online
Researchers at Censys found over 800 copies of Artex, a Chinese-made, open-source AI penetration testing tool, publicly accessible on the internet. The tool has been linked to several cyberattacks since July, and anyone could access the exposed servers. The finding highlights risks of powerful AI security tools being freely available.
1 new · 2 sources
Techmeme
TechCrunch