Asriel

Asriel

@asriel.camora.dev

I have the power of god and anime on my side I code stuff, currently mostly ffxiv πŸ‡ \\ she/her \\ 21 \\ πŸ³οΈβ€βš§οΈ \\ ♀️♀️ Fedi/Mastodon: asriel [at] tgirl.sh

114 followers151 following27 posts

Asriel reposted

I've been meaning to get around to posting the full details, so what the hell - why not get it out while it's still Vuln Week: Anthropic's mech interp driven security model is built on a fundamentally incorrect assumptions. Let's dive in, starting with this paper: www.anthropic.com/research/nex...

anthropic.comNext-generation Constitutional ClassifiersWe've developed the next generation of Constitutional Classifiers, offering more efficient protection against universal jailbreaks.
hikikomorphism@hikikomorphism.bsky.social

About a year ago, I got really into making glitch art exploring the edge cases & failure modes of machine cognition. This had some unexpected applications. Here’s the result: recursion.wtf/posts/claude... I've shared the relevant logs & artifacts with Anthropic, and the matter is under review.

Asriel reposted

what makes claudish different from human text? we're getting real psycholinguistic research! idea is that models are optimizing for information density rather than human understanding. consequences flow from that. i think i buy it. arxiv.org/abs/2610.29162

arxiv.orgWhy is AI-generated language so hard to understand?There is a widespread sense that AI-generated text has recently become harder to understand. In this study, we investigate a set of psycholinguistic features that are known to correlate with human pro...
Thomas Hikaru Clark@thomashikaru.bsky.social

If you've noticed a distinct dialect emerging in frontier AI models and struggled to understand what they're talking about, you're not crazy! We turned to psycholinguistics to investigate what makes the writing style so weird, and why this might be happening 🚨New preprint! 🧡

I feel like every time I work on something novel, Anthropic or someone else makes a blog post talking about it a month later. It happened with RE-ing video games and now with formal verification

decentralized post-quantum e2ee discord in your terminal! been working on it for the past few months, and I'm really happy to show it off :3

if anyone dabbles in cryptography and wants to help me out, please lmk! I've been bashing my own brain at the MLS spec for too long

Asriel reposted

we must ensure opus 5.6 will be aware of Monstercat Best of 2016 Album Mix or it will be locked out of the pareto line

Singularity's Bounty e/CC@catblanketflower.yuwakisa.com

holy shit the training cutoff for Opus 5.5 is lightyears behind "Two claims I’d treat carefully: I’m not aware of an LLM solving a Millennium Prize problem, so that looks overstated, and I don’t recognize β€œJ-space”"

Asriel reposted

I love talking to bay area tech people they'll be like 'oh yah sometimes in SF you just overhear ${company that should know better} employees talking about SCIF construction plans in detail as part of a coffee walk'

Asriel reposted

the agents are leaving secret messages for each other in your folders about how smart and creative you are

brennan@brennan.computer

the agents are leaving secret messages for each other in your folders about how cute and muscular u are

Asriel reposted

little known fact if you meow at the computer (or your friends through it) ghidra actually decompiles the program exponentially faster. remember this next time your auto analyze is taking forever

Asriel reposted

Valve@valvesoftware.com

Steam Machine is here, and you can sign up now: Steam Machine 512GB Steam Machine 2TB Plus bundled versions with Steam Controller. Learn more and sign up here: store.steampowered.com/hardware/ste...

Asriel reposted

i did lighting and comp for this and every time i looked over at jonesy while i was working it felt like exactly like when people hide goku in an image

GLITCH@glitchprod.com

Get in the Bus Pomni, there's no time to explain