That's the perfect way to describe the output. I find that the models tend to "overcomplicate" even the simplest of things. I think the next jump in their performance will come from distilling results down to "only the important stuff".

Wrote a new article for the Forward Future newsletter about Gina Raimondo's ridiculous belief that UBI would end America and her new extremely well-funded nonprofit built to spend $1 billion on job retraining pilots that don't work instead of basic income—despite its 200+ successful pilots.

Gina Raimondo Says Basic Income Would End America. Her New Billion-Dollar Org Is How She Plans to Stop It.

RAISE US bets on retraining as an alternative to UBI—but the evidence suggests AI displacement demands a bigger solution.

forwardfuture.com

Anthropic’s investors are expecting a $2 trillion IPO which is more than double OpenAI’s recent valuation of $850B when it let employees sell stock. It’s wild to think there’s at least a trillion in value that was created by the Claude Code team over the past year.

Anthropic investors bet on $2tn valuation in record IPO

Rapid revenue growth fuels hope Claude maker can overcome challenges to become the biggest listing in history

ft.com

I now have a real problem where when I see any writing with “it’s not just X, its Y” my brain just shuts down.

New class of supply chain attacks just dropped: split the malicious code across multiple packages. AI scanners inspect each package in isolation and see nothing suspicious. But the target project (Alibaba here) depends on all three, assembling the full attack. socket.dev/blog/npm-rat...

A US appeals court allows ~2,400 federal lawsuits against social media companies to proceed over claims their products were designed to addict young users (Diana Novak Jones/Reuters) Main Link | Techmeme Permalink

Making your website discoverable by AI is a must, but how do you actually know what LLMs care about? In this post, we share what LLMs read based on our internal experiments with roughly 268k agent-tracked requests to our site. Plus, a skill to apply these learnings:

Which AI actually reads your site? Two months of LLM traffic, measured—Martian Chronicles, Evil Martians’ team blog

We told you to measure. So of course we did it ourselves too: two months of server-side LLM traffic to evilmartians.com. Find out which agents read HTML, which read Markdown, what llms.txt actually does, and the one technique that got zero hits.

shorturl.at

The more I work in the @gleam.run compiler, the more impressed I am with all of the existing work it does. There is a lot of interesting stuff that goes into compiling pattern matching into efficient JavaScript.

I’ve changed the way controls work in hashi.giacomocavalieri.me, it would be nice if anyone who’s tried today puzzle could let me know if it felt better or clunkier 👉👈

I switched to Android and lost my favorite little video app. So I rebuilt the idea for the web: hold to record, arrange clips on a timeline, tap Go, and it exports one video to share. Everything stays on your device. Kody Video is live today: kody.video

Developers having unlimited AI tokens to use frontier models is like sales teams having unlimited business class flights. It never made any sense and every company will eventually have AI token budgets. The question will be how much and what’s approved to use the budget on?

Hugging Face just published a highly detailed technical account of OpenAI's accidental cyberattack on their systems - it's wild how sophisticated this was: huggingface.co/blog/agent-i... Wrote up some of my own notes here: simonwillison.net/2026/Jul/28/...

Anatomy of a Frontier Lab Agent Intrusion: A Technical Timeline of the July 2026 Incident

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

huggingface.co

LMFAO

npm is now scanning packages for malware at publish time before they become available for consumption

npm publish-time malware scanning and dual-use metadata - GitHub Changelog

As part of our ongoing supply-chain security work, npm is introducing automatic scanning of packages at publish time. This changelog covers what publishers can expect and a new metadata requirement…

github.blog

🚨 Announcing our new Podcast 🚨 Gus (@gworkman.bsky.social) and I have started a new #ElixirLang Podcast Macro Mayhem - @macromayhem.fm We cover Elixir news and general industry topics every 2 weeks. Our first episode is out now, available on all podcast players or on our website: macromayhem.fm

Macro Mayhem

The latest Elixir news and general software industry topics

macromayhem.fm

Tired of misleading cookie banners? The EU Commission has finally proposed a solution: set your privacy preferences in the browser once, and never see another banner. EFF joins a number of civil society organizations across Europe fighting to #KillTheCookieBanner killthecookiebanner.eu/

AI-assisted programming isn't that you sit back and relax. For building a production-ready product, you have to manage iterations, organize requirements, do QA, raise bugs to AI, and review code. You can't just vibe out a serious project without knowing what you're doing.

Update on this: My humble self and another very brilliant person will start a new Elixir podcast. First episode drops next week. Stay tuned.

Peter Ullrich@peterullrich.com · 2mo ago

I should start a podcast

The coverage on this OpenAI incident is abysmal. Use of the terms "rogue" and "loss of human control" lead to groupthink as people lack the critical skills to understand the difference between "autonomy" and faulty reward functions in AI on a task it was directed and given access to do.

I've been digging deep into what Elixir atproto OAuth can look like. This is not the final version yet, but it's starting to shape up. blog.annot.at/side-quest-a...

Side-quest: An Elixir atproto OAuth library

aka bikeshedding for great good

blog.annot.at

Stripe and a private equity firm teaming up to acquire PayPal was not on my bingo card. A pre-IPO startup acquiring a public company for $53B is the opposite of what we’ve come to expect. At this point I wonder if Stripe will ever IPO.

EXCLUSIVE: Stripe, Advent offer to buy PayPal for more than $53 billion, sources say

Payments company Stripe and private equity firm Advent International have made a joint offer to acquire ​PayPal Holdings Inc for $60.50 per share, in a deal that would value the ‌payments company at m...

reuters.com

I was a reviewer for this book so I can give you some spoilers about it: it rules, get a copy of it, start telling everyone about cumulative culture, unbuild the myths that keep your teams down, make shit be meaningful.

Cat Hicks@grimalkina.bsky.social · 2mo ago

IT'S POST'S BOOK BIRTHDAY! 🥳 Want a guidebook to the human side of tech? Want to cut through the noise of despair and anger to craft more meaningful work? Want to get sharp metacognitive strategies for this new era? Get all that *and* Launch Squad bragging rights here: www.drcathicks.com#book

Yesterday the Telegraph reported on my comments about air conditioning. The comment section is quite something. I have apparently become a hero of the anti net zero crowd by saying something what I think is sensible about air con. Allow me to point out three ironies in this 🧵

Dan Janowski breaks down the frontier of the Erlang ecosystem in his Code BEAM Europe talk - security, core components, compliance, and open source growth. The video is now public! Watch here: youtu.be/ctssvT9Cr4c #Erlang #BEAM

More evidence that men’s early deaths aren’t mainly about biology. They’re largely about culture (and about society treating men’s deaths as acceptable). In German monasteries, the gender mortality gap is far smaller than in Germany overall. www.cloisterstudy.eu/COMMS/

Soo this is rewriting the Erlang code generator so that it will be easier for us to switch from compiling to textual Erlang to the Erlang Abstract Format (EAF) binary format. That will have some nice benefits like (1/4)

New Elm release today! 🎉 0.19.2 🥳 Incremental builds of 850k lines of Elm in <350ms. Gearing up for a batch of smaller releases as we head towards Elm 1.0. elm-lang.org/news/faster-...

NIST is advancing nine new post-quantum signature algorithms as potential candidates for future standardization. We take a closer look at all of them, and argue that while they are in the works and show great potential, we should use ML-DSA for now. https://cfl.re/4ye4Uh2

Why we cannot wait for better post-quantum signature algorithms

Let’s have a look at the signature algorithms in detail. After that we’ll look at the timeline for their availability, and the reasons why we still need them.

blog.cloudflare.com

🌳 66 billion trees have been planted in China's Great Green Wall — and they appear to be growing faster than natural forests www.livescience.com/planet-earth... #science #climate #china

66 billion trees have been planted in China's Great Green Wall — and they appear to be growing faster than natural forests

A study of China's planted and natural forests reveals age, species mix, and CO2 sensitivity all contribute to how fast trees sprout leaves.

livescience.com

Right now, if you import something from JS, and the result is a 404 (or other not-ok status code), it fails the load & gives you nothing. I wonder with json imports, and maybe even text/byte imports, if that's the right pattern github.com/whatwg/html/...

Should modules really be 'null' if the HTTP status code is not-ok? · Issue #12657 · whatwg/html

What is the issue with the HTML Standard? https://html.spec.whatwg.org/multipage/webappapis.html#fetch-a-single-module-script:~:text=response%27s%20status%20is%20not%20an%20ok%20status%2C,-then%20s...

github.com