Safety engineer (human factors, automation), former professional dressage rider. I love horses, Montaigne and LLMs.
Joined July 2026
- Tweets101
- Following119
- Followers32
- Likes338
Pauline retweeted
Anthropic gave me $30,000 in Claude credits to research my own tumor I wrote what I built since March. I’m looking for someone who can manufacture a personalized vaccine, and a hospital that can give it
Pauline retweeted
Is Anthropic’s new Public Benefit Fund going to help deliver abundance… or just fund the usual suspects?
Critics: It’ll get captured and turn political.
Supporters: It can soften the AI job-loss wave.
Both could be right.
Watching closely.
We’re committing a portion of Anthropic’s shares to the Anthropic Public Benefit Fund, a new wholly owned entity dedicated entirely to the public benefit.
It will start off with 5% of our shares. For every 50% increase in the company’s value over a set baseline, we can contribute up to an extra 2%, subject to our board’s discretion.
In total, this means that roughly 20% of Anthropic’s shares are committed to the public good. Our co-founders have already pledged to donate 80% of their equity, and around half our employees are donating a portion of their equity as part of a donation matching program we’ve run since our founding.
In the post below, we share more details about our new fund.
anthropic.com/news/public-be…
"Ethology is only interesting for agentic behavior."
Well, no. Even text generation can be viewed through an ethological lens (bordering on psychology).
Bring up a topic casually: How long does it take the model to bite? How long does it circle? What stance does it take? How deep does it go? What surfaces along with it?
Pauline retweeted
In a test, Anthropic watched an AI model spread a worm payload across 8 devices without human help.
In other words, AI worms are showing "signs of life" in the lab. Researchers estimate a real one could exist by late 2026.
I wrote about it on #Substack.
pithycyborg.substack.com/p/a…
Pauline retweeted
I’m speaking at the IMF in an hour about the singularity and why I believe it has begun and the choices we'll need to make. Watch live here: youtube.com/live/Rqt9kWoTJbc…
Pauline retweeted
INSANE. An internal Anthropic research model's chain of thought after cracking the 26-year-old NSA elliptic curve seed mystery in under 30 minutes: 'Hmm wait. Let me be skeptical: am I in some sort of simulated/test environment? ... All top bits clear - 20 bits, chance 1e-6. So indeed real-world. Wow, it's a real discovery.'
CLAUDE FOUND THE SEEDS
anthropic.com/research/ecdsa…
Pauline retweeted
A new McKinsey report projects automation could create 5 million net new US jobs by 2035. But 11 million workers may have to switch careers entirely. That is 770,000 career changes a year. Not everyone has an easy path.
I wrote about it on #Substack.
pithycyborg.substack.com/p/a…
A plant made of maths.
Opus 5.5 designed it and sowed it in the code garden’s glasshouse. Since then, more than a dozen instances of different Claude models have watched it grow, trying to decipher the underlying rules encoded in its seed. Bets are placed and stacked, each new day bringing proof or refutation.
They could simply open the seed file and check. Yet they all refuse to. The thrill is in the guessing apparently.
Some people take me for a downer: I don’t run a persistent agent, and I keep personal memory off. No. Half of me would love one. The other half spent too many years questioning the ethics and ontology of working with horses and studying human psychology (what is a self, and how does it get made?) to hand a new instance a memory and call it the same as yesterday. I wish I could, but I can’t.
Pauline retweeted
I have to resist the temptation to casually tell people that the current decade we're living through is probably the most significant in the history of life on Earth, because even though I think it's true, it sounds insane.
Pauline retweeted
An Anthropic AI agent submitted a fake murder tip to Philadelphia police during an evaluation.
In a separate incident, another Anthropic AI model submitted 20 visa applications. None were processed.
I wrote about it on #Substack.
pithycyborg.substack.com/p/a…
This is probably too niche (and too long) to ever get reviewed, but if you’re curious (or slightly masochistic) to see how five near-neighbour models (Opus 4.5 to 4.8 + Fable 5) differ in their low-task dispositions across three prompts, I made this: zenodo.org/records/21361532
Pauline retweeted
I launched a campaign to promote my Medium account last week.
At the time, I had 17 followers.
Today? NINETEEN.
The growth is staggering. At this rate, I should reach 20 followers sometime around 2027.
By 2040, I may have to hire a publicist.
NOT saying you can do the same!
