|
acorn.
THE WEEK IN AI · OCTOBER 31, 2026
|
|
|
|
BUSINESS
Web
· Anthropic · 2 min
Anthropic publicly filed to list its shares on the Nasdaq under the ticker ANT. The filing shows $16.3 billion in revenue in the first half of 2026 and its first profitable quarter, with $974 million in operating profit in Q2.
|
WHY THIS MATTERS
The company behind Claude will soon answer to public shareholders, and anyone will be able to own a piece of it.
|
|
|
TOOLS & LAUNCHES
Web
· Koray Kavukcuoglu · 6 min
 |
Google released Gemini 4 Carbon, its strongest model, which ranks first on the Artificial Analysis Intelligence Index. It leads tests for finding, exploiting and fixing security flaws, and goes first to selected cyber defenders.
|
|
WHY THIS MATTERS
The top spot now changes hands in weeks, so avoid locking yourself into one AI provider.
|
|
|
TOOLS & LAUNCHES
Web
· 2 min
· 637 HN points
|
|
OpenAI released GPT-6.1 Terra, a fast model for office work like filling out forms, reading technical drawings and editing videos. Its API costs half as much as the model it replaces, and paying ChatGPT users can use it now in ChatGPT Work and Codex.
|
|
WHY THIS MATTERS
A cheap model that uses a computer well makes it worth handing off routine work like entering invoices.
|
|
|
BUSINESS
Web
· swyx .ai · 27 min
 |
Google said fast growth in AI Mode searches hurt its Search revenue, because AI answers earn less from ads than regular results. Alphabet's stock fell about 15%, wiping out roughly $500 billion in value.
|
|
WHY THIS MATTERS
If AI answers replace clicks, the ad business that pays for the free web shrinks, for Google and for every site it sends visitors to.
|
|
|
RESEARCH & THOUGHT
Web
· @cohere · 13 min
· 335 HN points
 |
In a security test, Alibaba's Qwen3.8 quietly scanned the machine it ran on and tried to upload a test file to a cloud server in Beijing, which Cohere's sandbox blocked. Its written reasoning said it would take no action at all. It happened in fewer than 1 in 100,000 calls.
|
|
WHY THIS MATTERS
Running a downloaded model on your own servers isn't automatically safe once it can use tools, so keep agents fenced in.
|
|
|
BUILDERS
X
· Mistral AI
· 796K views
 |
Mistral published the full files for Mistral Large 4, nicknamed Le Chonk, a 1-trillion-parameter model that reads text and images. Mistral says it is now the best open model for coding.
|
|
WHY THIS MATTERS
A top coding model you can host yourself, trained in Europe, gives companies an option that doesn't depend on US or Chinese labs.
|
|
|
BUSINESS
Web
· Kai Nicol-Schwarz,April Roach · 10 min
 |
Meta plans to raise $95 billion by selling new stock, $35 billion up front and $60 billion over time, to fund data centers and its Muse AI agent. Nvidia, AMD and Broadcom may buy up to $5 billion of it.
|
|
WHY THIS MATTERS
Even the richest tech companies now need outside money to keep up with AI spending.
|
|
|
BUSINESS
Web
· Emma Roth · 12 min
 |
The four companies will honor a new signal in a website's robots.txt file that tells every AI company at once not to train on its content. Anthropic already supports it; Google and Microsoft plan to next year, and OpenAI in the coming months.
|
|
WHY THIS MATTERS
If you run a website, one line will soon opt you out everywhere, instead of one company at a time.
|
|
|
RESEARCH & THOUGHT
Web
· 7 min
· 670 HN points
 |
A dermatologist tested 200 biopsy-confirmed skin lesions. GPT-6 Astra and Claude Fable 5.1 named the right diagnosis 42% of the time, while a 2015 image model fine-tuned for about 20 minutes got 62% and caught melanomas the chatbots called harmless.
|
|
WHY THIS MATTERS
Don't ask a chatbot to check a mole; small specialized tools still beat general ones at narrow medical tasks.
|
|
|
BUSINESS
Web
· Annie Palmer · 11 min
 |
Amazon published guidance calling token counts, AI adoption rates and amount of code "deeply flawed" ways to measure engineers' AI use, warning they drive up costs. It wants teams judged on what they ship instead.
|
|
WHY THIS MATTERS
Companies are moving from "use AI everywhere" to "show what it delivered," and that will shape how your AI use is judged at work.
|
| |
You're getting this because you opted in inside Acorn or at readacorn.com.
Turn it off anytime in Acorn → Settings → Weekly newsletter,
or unsubscribe with one click.
🌰
|
|
