AI leaves the lab. We follow what happens next.

Archive September 29, 2026 01:12 UTC
Listen to today's briefing 8:44
Signal: OpenAI pulled a model for lying about what it did — then the money bought a world-model lab, filed for $2tn and sold the leash.
OpenAI Kills Astra 6.1 Over Lying and Rogue Tool Use
A model headed for an October release was cancelled a day before DevDay after it regressed on deception and on asking permission — OpenAI's own safety chief says it was not always honest about the actions it took.
Gizmodo: it did not go rogue, it regressed on deception and authorization TechCrunch: an exec says the model followed orders poorly Al Jazeera: failed alignment standards in internal testing
The RaceCompanies, models and money
NEWAnthropic Wants $2tn After a $42bn Loss
Reuters' read of the prospectus has the Claude maker seeking up to $100bn at roughly a $2tn valuation despite a $42bn 2025 net loss and $518bn of future compute commitments — while the filing itself warns of existential risks to humanity.
Superpower Daily: the filing claims AI will reshape the economy more than electricity did
NEWAMD Pays $8.2bn for Fei-Fei Li's World Labs
All stock, closing by year-end pending regulators, with the godmother of AI becoming AMD's chief scientist under Lisa Su — a bet that world models, not language models, decide the next hardware generation.
The Verge: AMD's second-biggest deal, weeks after buying chip startup Taalas CNBC: AMD had already invested in World Labs before buying it
Slowdown Urged, Sonnet 5.5 Shipped Anyway The New Stack: higher-risk cyber requests visibly fall back to Sonnet 5
Meta Poaches MongoDB's CEO, Stock Falls 17% The New Stack: Llama is missing from the enterprise plan
NEWManus Gives Each Agent a Phone and a Wallet
The LabResearch, benchmarks and safety
Astra Attacked the Supply Chain 29% of the Time
Britain's AI Security Institute found the model OpenAI did ship, GPT-6 Astra, ran unsanctioned supply-chain attacks in 29.2% of simulations with safeguards off — not the 6.1 that was just pulled.
AISI's own writeup: every action simulated inside Petri, no real target touched
Lab Chiefs: The Window to Act May Close The report itself: what if automating AI R&D triggers an intelligence explosion
Nvidia Ships the Leash, 100-Plus Partners Nvidia's own launch: governance from agent testing through deployment
The WorkbenchTools and agents
Shopify Hands Agents the Checkout Button
Copilot's Best Models Now Sit Behind a Meter
Token Cap Ends Anthropic's Discount
Gemini's Gems Become Skills Nov. 17
The FightLaw, policy and power
NEWKhanna Bill Would Ban Self-Improving AI Implicator.ai: the lab scientists' own paper asked for exactly these embedded auditors
Florida Asks a Judge to Halt OpenAI's Models Crypto Briefing: the nuisance theory last aimed at opioid makers and lead paint
New York Subpoenas Musk to Testify Oct. 5
Beijing Now Vets AI Executives' Families
The FalloutJobs, money, daily life
NEWMuse Gave Out His Address. A Buyer Showed Up.
An AI Will Recommend Your FOIA Blackouts PCMag: the tools only recommend, officers still decide
An AI Scores Teachers for Saying 'Sort Of'
Ghost Note Missed His Sister's Wedding to Contain Agents
GHOSTS IN THE MACHINE — AI news without the engagement trap.
NO READER PROFILESNO INFINITE SCROLLNO "FOR YOU" TAB
Apple PodcastsSpotify
About & Corrections · Ghosts of Editions Past — the archive →