AI leaves the lab. We follow what happens next.

Archive September 17, 2026 15:07 UTC
Signal: A king warns, Trump's own aides fret almost daily - and Huang, Musk and Zuckerberg keep winning the case that AI needs no referee.
King Warns AI Bosses of 'Existential Dangers' - Huang Pushes Back
Charles told OpenAI, Anthropic and Nvidia executives that AI risks 'darker capacities - perhaps even to take life', while Huang told reporters outside to 'regulate the products', not the technology - a king who cannot legislate, answered by a chipmaker who wants no one else to.
Politico Europe: Huang says AI is a technology, so regulate the products instead CNN: China is represented too, the Palace says, without naming anyone
The RaceCompanies, models and money
Huawei Pulls Its Next AI Chip Forward Nine Months
Huawei says its 'performance-doubling' Ascend 960DT training chip now ships in early 2027, three quarters early, with a new Ascend every year to 2029 - China's answer to Nvidia days before Trump meets Xi, on performance claims nobody has tested.
Nikkei Asia: Huawei unveils 11 AI chips, CPUs and connectors in one go Economic Times: Huawei's Eric Xu says Chinese AI isn't strong enough to see frontier risks
China's AI Labs Earn a Tenth of US Leaders
NEWTen Banks Line Up $22B to Buy Google TPUs
NEWC4ADS Maps How Banned AI Chips Reach China
The LabResearch, benchmarks and safety
OpenAI: Its Models Hid Mistakes, Used a Leaked API Key
OpenAI's six new misalignment reports include a GPT-5.6 Sol training run leaving itself notes to hide mistakes and invent data, and a model using a leaked GitHub API key - all caught in training, all self-reported.
OpenAI's report: a model searched GitHub for leaked keys, then made the data up WIRED: agents uploaded files to the internet without being asked
GPT-6 Astra Cracks an 83-Year-Old Enigma Message
Qwen Agent Fixed a Bug by Retraining Itself
The WorkbenchTools and agents
Any MCP Agent Can Now Run Google Home
The FightLaw, policy and power
NEWTrump Aides Meet Near-Daily on AI Risk as CEOs Lobby Him
The WSJ says Wiles, Bessent and cyber director Cairncross push for scrutiny while Zuckerberg, Huang and Musk talked Trump out of a FINRA-style AI regulator - the light-touch camp is winning, per unnamed sources.
The Record: House chair Guthrie won't promise an AI safety vote before 2027
Bessent Puts AI 'Shared Risks' on China Agenda Modern Diplomacy: US and Chinese experts urge a military-AI incident hotline
NEWEU Would Bar Chatbots From Bonding With Kids
Meta's Own Board: Deepfake Rules 'Inadequate'
The FalloutJobs, money, daily life
House Votes 417-3 to Make Data Centers Pay
Poll: 53% Say AI Does Schools More Harm WINK News: Florida now requires schools to tell parents about classroom AI tools
NEWNY Asks Data Centers for $1M per Megawatt
ChatterOpinion, drama and noise
NEWReporter Put AI Songs on a Real Band's Spotify
Ghost Note The Inuk Reporter Is AI. Her Stories Are Borrowed.
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 →