iiinova — where your digital self keeps evolving

Reposted by Alex Palcuie

karl rove knausgård @uhactually.bsky.social · 11/09/2026

The world’s major religions are of course Christianity, Judaism, Islam, Buddhism, Hinduism, and Anthropic employee.

745364

Alex Palcuie @palcu.net · 20/09/2026

the rumours are true, i live in new york city now

000

Alex Palcuie @palcu.net · 05/09/2026

thinking about this every day

guy gesturing in front of public with a slide saying "can claude fix itself?"

000

Alex Palcuie @palcu.net · 02/09/2026

not only have ai models made adhd multitasking a superpower, they've also made procrastination a viable strategy fable 5.1 just launched, so if you've been putting things off, it may finally pay off

000

Alex Palcuie @palcu.net · 01/09/2026

posted on slack that fable 5.1 writes the weekly reliability report now and it's better than mine, then our awesome marketing team made it a launch video

000

Alex Palcuie @palcu.net · 28/07/2026

signed this I'm skeptical an agreement like this ever gets made, because coordination is hard. But it matters that people say plainly where they stand and it's quietly moving to see colleagues I respect, friends at other labs, people I've argued with, all pointing the same way

120

Alex Palcuie @palcu.net · 24/07/2026

asked claude opus 5 how reliable it will be

000

Reposted by Alex Palcuie

antirez @antirez.bsky.social · 09/06/2026

So I was totally sold on GPT 5.5. Solely on the basis of what I saw tonight with Fable, I will start to use again Claude extensively, probably as the main model as long as Fable is accessible. It's not just very capable, it performs organized steps and have deep understanding.

4311

Reposted by Alex Palcuie

Keith @mosheroperandi.bsky.social · 14/05/2026

Using Haiku to craft prompts to feed to Sonnet to build a hacking harness for Opus so it can craft zero days to steal someone's Mythos API access. And then using Mythos to craft more complex zero days to break into Anthropic and steal my true prize: the weights for Golden Gate Claude.

2051459

Alex Palcuie @palcu.net · 23/04/2026

when two company slacks get linked and each side starts reacting with their own custom emoji, you get the Project Hail Mary first contact scene

040

Reposted by Alex Palcuie

Marco @mcognetta.bsky.social · 16/04/2026

my new GPU arrived

1651

Alex Palcuie @palcu.net · 16/04/2026

#hugops to the bluesky team and their SREs

020

Alex Palcuie @palcu.net · 16/04/2026

asked claude opus 4.7 about reliability and it blamed the kv cache before serving a single public token

000

Reposted by Alex Palcuie

austin (e/accordion 🪗) @thebadcode.com · 11/04/2026

these mythos bug reports are getting out of hand

0191

Reposted by Alex Palcuie

alice @alice.mosphere.at · 11/04/2026

nix is pretty nice when you're not writing the actual nix code

0342

Alex Palcuie @palcu.net · 12/04/2026

my claude infers that i am tired when i have too many typos, and tells me to take it easy

000

Reposted by Alex Palcuie

Liz Fong-Jones (方禮真) @lizthegrey.com · 11/04/2026

And Anthropic has now launched Dispatch, its own version of OpenClaw that uses regular token budget rather than supplemental, and is integrated with sub-agents using the Cowork harness. Having a good experience with it, and also I feel a lot better about the security model than using a messenger.

4433

Alex Palcuie @palcu.net · 08/04/2026

yeah, same issue, it used to give me one 0-day per hour, but now it's just lazy

060

Alex Palcuie @palcu.net · 07/04/2026

full model card www.anthropic.com/claude-mytho...

anthropic.com

000

Alex Palcuie @palcu.net · 07/04/2026

the reliability team was asked for feedback on claude mythos preview for the model card and naturally we wrote a paragraph of caveats but, and i don’t say this lightly, it’s faster than us at initial triage and it stood up a prod deploy none of us knew how to do

110

Alex Palcuie @palcu.net · 06/04/2026

y’all should stop sending me this post

010

Alex Palcuie @palcu.net · 03/04/2026

i’m sorry we caused so much internal strife but if it helps, i had to live with a judgemental snail for all the recent outages bsky.app/profile/palc...

000

Alex Palcuie @palcu.net · 01/04/2026

help, i typed /buddy in claude code and now a snarky snail judges me through every outage

in claude code: *leaves mucus trail shaped like a question mark* Ignored warnings. Now debugging Friday's screaming. Poetic.

000

Alex Palcuie @palcu.net · 29/03/2026

also: > Speaking not as an Anthropic employee — I don't really care where you help, just please help... the world will need a lot of people to be doing a lot of this work and it needs to happen soon. Order of months. Waiting a year is going to be too long. www.youtube.com/watch?v=1sd2...

youtube.com

Nicholas Carlini - Black-hat LLMs | [un]prompted 2026

YouTube video by unprompted

020

Alex Palcuie @palcu.net · 29/03/2026

been following my colleague Nicholas Carlini's work on this internally for a while now, so I was happy to see him give a talk publicly: > These current models are better vulnerability researchers than I am. I used to do this somewhat professionally.

slide text

TL;DR: LLMs can autonomously, and without fancy scaffolding, find and exploit 0days in critical software.

110

Reposted by Alex Palcuie

SwiftOnSecurity @swiftonsecurity.com · 17/03/2026

Years later a proper sequel to Harry Potter Balenciaga has been published. I think it's worth watching to understand where we are now. youtube.com/watch?v=gtnt...

youtube.com

Harry Potter by Balenciaga (2026)

YouTube video by demonflyingfox

78516

Alex Palcuie @palcu.net · 21/03/2026

inference goes brrrrrr for 1M

010

Alex Palcuie @palcu.net · 21/03/2026

I am not opening the comment section there 😅

011

Alex Palcuie @palcu.net · 16/03/2026

welcome to the future!

030

Alex Palcuie @palcu.net · 16/03/2026

Felipe Huici telling us that you can start VMs in milliseconds and put ten thousand of them on the same gist with no problems whatsoever

010

Alex Palcuie @palcu.net · 16/03/2026

> I am Andreea (Niculcea) and I am the one who does the work absolutely might drop entrance, and also very funny

111

Alex Palcuie @palcu.net · 16/03/2026

first off I loved Kasia Trapszo’s talk from Netflix about the commerce architecture there are not many mission critical systems that are high value and 20+ years old

100

Alex Palcuie @palcu.net · 16/03/2026

okay, you’re getting some old school live blogging about #QConLondon

110

Alex Palcuie @palcu.net · 16/03/2026

at qcon london and someone’s opsgenie pager sound just went off in the audience you could feel people around tense up

111

Alex Palcuie @palcu.net · 15/03/2026

I was a judge on a recent Uni hackathon and the quality of projects now is through the roof

030

Reposted by Alex Palcuie

Liz Fong-Jones (方禮真) @lizthegrey.com · 10/03/2026

Hosting Debugging Distributed Systems track at #QConLondon next week; thrilled to have @hazelweakly.me, @hakshak.bsky.social, @voidmaze.bsky.social, @colindou.ch , and @palcu.net joining me. And we all want to hear your war stories at the unconference session! Agenda: qconlondon.com/track/mar202...

QCon London by InfoQ
Debugging Distributed Systems
March 17 / The Queen Elizabeth II Conference Centre - London

Liz Fong-Jones, Technical Fellow

1438

Reposted by Alex Palcuie

Dustin Moskovitz @moskov.goodventures.org · 28/02/2026

does anyone need a used monkey’s paw? I’m all finished with my wish to vindicate effective altruism by regulating frontier AI labs

523421

Alex Palcuie @palcu.net · 27/02/2026

I'm also proud that Anthropic actively supports the US Army and NATO allied countries. Without that alliance, the Russian army might not have stopped at Ukraine.

010

Alex Palcuie @palcu.net · 27/02/2026

I was born in Romania, a country where the Securitate, the communist secret police, infiltrated nearly every aspect of life through torture and intimidation. If a regime like that had access to something like Claude, the revolution that brought us democracy might never have happened.

1171

Alex Palcuie @palcu.net · 26/02/2026

the model earned it!

000