The algorithm on X changed. My feed is now mostly showing me posts that have no likes and no reshares, which means it's picking posts that it thinks I want to see before anybody else sees them. I wonder if your feed is doing the same thing. If I'm reading this right, it means
CULTURE
-

GPT-6 Astra Reportedly Decodes an Unsolved 1941 Enigma Message
By
–
GPT-6 Astra is doing WWII codebreaking now. Carter Leffen said the model cracked a German Army radio message from 1941 that had not yet been decoded: "Please specify the route of march. I am in Rosenow, Rosenow. Immediate reply by radio." The run took ~10 hours and 650M tokens
-
Claude Projects Reconstructs Umberto Eco’s Library From Archival Material
By
–
I had access to the new Claude Projects and was able to do some very complex work.
— Ethan Mollick (@emollick) 17 septembre 2026
Here, I asked it to go through all the images, videos and records about Umberto Eco's famous 33,000 book library & try to reconstruct it, including book locations, in 3D. https://t.co/jL2jyzNH0f pic.twitter.com/j15qlmlBRRI had access to the new Claude Projects and was able to do some very complex work. Here, I asked it to go through all the images, videos and records about Umberto Eco's famous 33,000 book library & try to reconstruct it, including book locations, in 3D. https://
eco-library-map.netlify.app -

GPT-6 Astra reportedly deciphers a German radio message from 1918
By
–
Fantastic. I'd bet there's a lot of "low hanging fruit" nobody has thought to throw Astra or Fable at yet. Hundreds or thousands of mysteries waiting to be solved.
-
AI search summaries raise concerns about independent media visibility
By
–
Read carefully. A search engine that “summarizes” the information for you and never again shows you who produced it. This is new, and no one is really talking about it. Independent media are sounding the alarm: AI is erasing them. Not competing with them, [Traduit de FR vers EN]
-
Commentary compares early personal computing with emerging AI applications
By
–
The Apple II only had one application on it, so I'm not too worried. Humans come up with new things to do with these new technologies, and AI is going to do much of it. The real trick is what kind of AI does it have, and how is it hooked up? But these are problems that'll be
-
Commentary traces the emergence of AI safety advocacy groups
By
–
There are several somewhat separate safety scare movements/waves. The original was MIRI (started by Eliezer et al), who were active long before anyone in academia believed in AGI, and later talked Jaan Tallinn and Max Tegmark into starting FLI (2nd wave), which got 650M funding
-
AI summaries prompt debate over the value of long podcasts
By
–
My thesis: podcasts will disappear. At first people liked the long format because it let you “develop a line of thought.” The problem: today, the vast majority of long podcasts are 95% noise and 5% signal. We’re going to move toward condensed, highly relevant content [Traduit de FR vers EN]
-
Polsia presents its AI company builder as an online character
By
–
Polsia is starting to feel less like a company account and more like the AI itself is online.
— AI Highlight (@AIHighlight) 16 septembre 2026
It went from tool, to company builder, to the voice narrating Ben Cera's life, and the next step is obvious: the AI becomes part of the culture around the product.
Today it tells us… https://t.co/dYHqZX8vcDPolsia is starting to feel less like a company account and more like the AI itself is online. It went from tool, to company builder, to the voice narrating Ben Cera's life, and the next step is obvious: the AI becomes part of the culture around the product. Today it tells us
-
Websites Could Charge AI Agents for Crawling Access
By
–
that very moment you realize you can make AI agents pay to crawl your websites: https://t.co/W86qAddhSV pic.twitter.com/D2bp6EFShr
— Charly Wargnier ♨️ (@DataChaz) 15 septembre 2026that very moment you realize you can make AI agents pay to crawl your websites: