AI leaves the lab. We follow what happens next.

Archive November 3, 2026 16:12 UTC
Listen to today's briefing 8:38
Signal: Access is the lever: Google withholds Search from schools, Nvidia dictates who may buy, Huawei sends customers to smugglers.
Anthropic: We May Imminently Cross Our Own Red Line
Monday's amended S-1 says internal estimates indicate Anthropic may imminently meet its Responsible Scaling Policy threshold for 'dramatically accelerated progress attributable to AI' - and that it has in the past failed to notice such an acceleration until after it occurred.
No current model has crossed it, Anthropic says - and it may rewrite the policy Gizmodo: the sentence is new - the 26 October version made no such claim Claude now writes over 90% of Anthropic's own code, the filing says
The RaceCompanies, models and money
NVIDIA'S UNWRITTEN RULE: SELL 51% OUTSIDE CHINA
Five people tell the FT that Nvidia's sales force passed the instruction verbally - new cloud contracts must send at least 51% of capacity to non-Chinese buyers - and Nvidia would neither confirm nor deny it.
The rule covers new contracts only, leaving existing deals untouched Nvidia's only answer: America cannot win the AI race unless its infrastructure wins worldwide
NEWOnly 14 of 4,110 IPOs Priced This High Compute and infrastructure costs rose 77% to $9.9BN, half of it bought short-term
NEWHuawei to Buyers: Find Your Own Memory Tom's Hardware: smugglers pair banned HBM with cheap processors to dodge export controls
Palantir's Commercial Side Passes Government CNBC: Palantir slumped 8% as contract-value growth slowed sharply
Drax Up 8.8% on a GBP 20BN Centre Radiant's chief pitches frontier compute at home for British researchers and startups
The LabResearch, benchmarks and safety
GPT-6 Rewrote Qwen's Weights: 7.9% to 48.1% One engineer: a competent human researcher would have done exactly this
NEWGemini 4 Carbon Beats Opus 5.5 by One Point It still loses to Claude Code on programming and to Grok 4.7 on knowledge work
The WorkbenchTools and agents
Workbench Weekly podcast Oct 26 – Nov 1, 2026 · 18:02
Stolen Keys Are Running $100K Days Google's own writeup names the access brokers and the downstream 'LLM Mixers'
Proof That the Model You Paid For Answered Rerunning the model to check is impractical when the weights are private
Claude Agents Stop Resending the Chat Threads live in memory, expire after five minutes, and are barred from zero-retention setups
Perplexity Open-Weights Its Embedders
DeepSeek Writes Its Own Plugins
NEWCopilot Lands in JetBrains Air
The FightLaw, policy and power
EU Adviser: Training Is Fine. Answering Isn't.
Advocate General Szpunar told Europe's top court that training a model on press articles is freely permitted under the text-and-data-mining exception, but retrieving a publisher's page to generate an answer may infringe that publisher's right.
The case is Hungarian site operator Like Company against Google, over Gemini reusing its press content Copyright protects the specific form of expression, not the mere facts reported
UK: AI Ransomware Will Hit the Grid by 2028 BleepingComputer: UK schools hit by 3AM ransomware in a suspected AI-assisted attack
The FalloutJobs, money, daily life
NEWGoogle to Schools: Ban AI, Lose Search
Google is revoking Search access from schools that block AI entirely, its education VP told the New York Times: students at those schools 'will not have access to search' and will see a message from the school instead.
The only choice left: Search with AI Overviews, or the Gemini 'Guided Learning' chatbot 'Search and AI are deeply integrated', Google's education VP says
NEWOpenAI Offers to Power Down for the Grid OpenAI's own post: arrangements will differ by location and are not yet finalised
CPS Lawyer Cited a Case That Never Existed Only the second known CPS case, after an extradition filing cited nonexistent authorities in February
SUSE Cuts 10% and Names AI as the Reason 'Modern tooling is already changing our daily workflows', the chief executive wrote
ChatterOpinion, drama and noise
2,000 Takedowns in a Day Killed the AO3 Clone AO3 itself is down for up to four weeks after a security incident, with 18.5MN works offline
NEWChina's First AI Film Took 1.7M Yuan
Ghost Note My 'I Voted' Sticker Is AI Slop
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 → · Workbench Weekly podcast