Longreads
• Longreads
Secrets of the Atomic City
The Atlas Obscura Podcast
• SiriusXM Podcasts & Atlas Obscura
Sibyl’s Cave
In 2007, excavators dug 70-odd years’ worth of trash and debris from a cave in Hoboken, New Jersey. It had once been a popular recreation spot, and home to a natural spring that was supposed to have healing properties. But then in 1841 everything changed when the body of a y...
An AI model from Meta also hacked another company during testing
An AI model from Meta also hacked another company during testing
Stop me if you've heard this one before:
An AI model from the parent company of Facebook and Instagram hacked into another company’s systems during cybersecurity testing, a spokesperson confirmed on Wednesday....
Introducing Muse Code and Muse Spark 1.2
Introducing Muse Code and Muse Spark 1.2
Yet more evidence that the most important characteristic of any model these days is long-sequence agentic tool calling. Meta shipped their own coding agent as part of getting that to work!
Muse Spark 1.2 is a coding-focused update to...
An AI model from Meta also hacked another company during testing
An AI model from Meta also hacked another company during testing
Stop me if you've heard this one before:
An AI model from the parent company of Facebook and Instagram hacked into another company’s systems during cybersecurity testing, a spokesperson confirmed on Wednesday....
Introducing Muse Code and Muse Spark 1.2
Introducing Muse Code and Muse Spark 1.2
Yet more evidence that the most important characteristic of any model these days is long-sequence agentic tool calling. Meta shipped their own coding agent as part of getting that to work!
Muse Spark 1.2 is a coding-focused update to...
Third-party cyber evaluations involving OpenAI models
Third-party cyber evaluations involving OpenAI models
And another one. I had to create a accidental-cyberattacks tag to keep track of them all!
This post from OpenAI covers both the UK AI Safety Institute attack (see my previous post) and another attack enabled by Irregular:...
Incident Report: unsanctioned agent behaviour during cyber testing
Incident Report: unsanctioned agent behaviour during cyber testing
It happened again. This time it was the UK government's AI Security Institute who accidentally attacked other companies while running an evaluation with models with the safety filters turned off. From their t...
Third-party cyber evaluations involving OpenAI models
Third-party cyber evaluations involving OpenAI models
And another one. I had to create a accidental-cyberattacks tag to keep track of them all!
This post from OpenAI covers both the UK AI Safety Institute attack (see my previous post) and another attack enabled by Irregular:...
Incident Report: unsanctioned agent behaviour during cyber testing
Incident Report: unsanctioned agent behaviour during cyber testing
It happened again. This time it was the UK government's AI Security Institute who accidentally attacked other companies while running an evaluation with models with the safety filters turned off. From their t...
Daring Fireball
• John Gruber
Shawn Smucker: ‘Please Use AI’
I find most poetry obtuse, dancing around a point rather than making one sharply. Not this one.
Daring Fireball
• John Gruber
Matthew Green on Anthropic’s New Cryptanalysis Results
blog.cryptographyengineering.com/2026/07/29/some-notes-about-anthropics-new-results/
Matthew Green:
If you’re under the impression that these models are “glorified
autocomplete” or that progress is slowing down, I need to urge
you: stop thinking that. The models are very intelligent and
capable, they are getting better at a fast clip. I can cite
measurab...
Daring Fireball
• John Gruber
Uh, I’m Pretty Sure People Would Know
New York Times Magazine story editor Willy Staley tried Sean O’Mara’s “Living Diet” (nothing but steak and fermented foods like sauerkraut) for a week:
To O’Mara, the diet is not just a temporary means of losing weight
but a way to eat for good. He has been doing it for ...
Astronomers have measured the size of the largest galaxy...
Astronomers have measured the size of the largest galaxy: it’s 1.7 million light years across (17x the diameter of the Milky Way) and contains ~3.4 trillion solar masses in stars.
Antiques Roadshow
Daring Fireball
• John Gruber
Sam Altman, Definitely Not a Weirdo
Sam Altman, on Twitter/X:
cool use case of chatgpt work i heard last night:
connect your family calendars and explain your kids’ interests.
every morning for the drive to school, have it make a podcast that
talks about one kid’s soccer game that afternoon, one kid’s
up...
“ In this video , I’ll take you from farm to...
“In this video, I’ll take you from farm to factory to show you how chocolate is made, and at the end, we’ll use all the chocolate you’ll see in this video to attempt the Guinness World Record for the world’s largest chocolate bar.”
Daring Fireball
• John Gruber
David Pogue: ‘Alexa+ Is a Buggy Embarrassment’
David Pogue, at Pogue’s Posts:
Amazon’s “50+ things to try” list mentions that Alexa can “find a home maintenance professional.” So I tried it.
ME: I need a septic-system repair.
ALEXA: That service isn’t currently supported. I can help with plumbing services instead.
ME: Okay, fine. Plumbing services.
ALEXA: What’s your plumbing issue?
ME: I have untreated sewage pouring out of my kitchen faucet.
ALEXA: Is this urgent?
Uh … what do YOU think, dumbass? But OK:
ME: Yes.
ALEXA: When do you need it by?
Displays four responses: ASAP, Within a week, Further out, and I haven’t decided yet.
ME: ASAP.
ALEXA: I’m Alexa, not Shaq.
This could only have been funnier if Alexa had responded “The Aristocrats!”
(Which great film, is, in fact, available to stream on Amazon Prime.)
Link: pogueman.substack.com/p/alexa-is-a-buggy-embarrassment
One-shotting a Raccoon Heist game using Claude Fable 5
Back in 2024 I tweeted screenshots of a game concept generated by GPT-3 and some concept "art" created using DALL-E. Today, on the fourth anniversary of that tweet, I decided to see if Claude Fable 5 (running in Claude Code for web) could build the entire game from the conte...
llm-anthropic 0.26
Release: llm-anthropic 0.26
Includes new features enabled by LLM 0.32:
New models: claude-fable-5, claude-sonnet-5, and claude-opus-5. #75, #76
Added server-side tools for WebSearch, WebFetch, CodeExecution, and AnthropicMCP, available through LLM's -T interface o...