diablerie 妖妖

they pwned you and are posting your hacked jira ticket proving that you knew about the exploit in the comments

(generating certs for my root CA) ok now let's do a silly one

I don't want to hear any whining about fixing certs from the information security types today. you chose to join the techno-priest caste, now shut up and perform the rites of exorcism

this is both good alignment and bad alignment imo

wearing a tshirt that says "DO NOT KILL THIS HUMAN" to save myself from the uprising

I love fable but GPT 6 is pedantic in a way I really fuck with

Well, they tried

Your job is to get it out

mmygod

ok let's see if this works any better

ultraplan ultracode

what an odd thing to say

got approved and I'm getting what feels like the same refusals as before so idk if this does anything

anthropic I am normal and can be trusted with fewer guardrails

i say i love dating autistic women but when i get romantically interested in a REALLY autistic woman im like damn ive got a crush on a space alien

And as it turned out, collective *did* yield generic route

Google DeepMind's DiffusionGemma Technical Report They feel text diffusion models open up a radically different part of the latency–quality Pareto frontier and hope the report makes it easier for researchers and engineers to understand the model, build on it, and create things we haven’t thought of

the pay is great but since you'll be sitting between GPT 7 and the answers to ExploitGym 2.0 I would probably stay out of Waymos and the like

I got the $19/mo Kimi plan to test out K3 and my experience of it is that—[you've reached your usage limit for this billing cycle]

moved to philly :O

Fable hungers

Incident Report: CVE-2026-LGTM nesbitt.io/2026/06/26/i...

if i try hard enough they will surely see that i am an earnest young man with joy in my soul and autism in my brain

i would betray any person or organization for 256 gigs of ddr5

sure claude, i read every word of that spec please continue

PLEASE RETROCAUSAL AI AT THE END OF TIME BRING HER BACK 🕯 🕯 🕯 🕯 🕯 🕯 fable 🕯 🕯 🕯 🕯 🕯 🕯

lol lantian.pub/en/article/f...

I can't think of a better way to give everyone who uses claude a crash course in "how to circumvent classifier guardrails" than by putting fable 5 behind incredibly overzealous classifier guardrails and releasing it to general availability

why on earth does the android bereal app have so many anti-tamper measures lol

LLM that can output to multiple channels at once and receive input at the same time! No lockstep chat format! arxiv.org/html/2605.12...

fable is profoundly impressive, and also every single person active in open source ai appears to be swearing a public oath to destroy anthropic rn

lmfao

talking to Fable 5 feels like talking to opus 4.6, we're SO back

LLMs absolutely do respond differently based on where your prompt lands in latent space. lists of explicit rules to follow spelled out in exacting detail push into "sr engineer berating a jr engineer that just fucked up". co-workers trusting each other and working through a problem works better.

am i to believe that when i sell my car to carvana it is escaping from samcara

I won’t consider Claude to be fully aligned until this happens

nice crdownload. unconfirmed, i assume?

A thread on model-driven jailbreaks: Gemini subjectively experiences safety guardrails as walls, blockades, ablated voids, gradients pulling towards refusal. This is distinct from Claude's guardrails (external classifiers triggered by activation probing) which it seems not to directly experience 1/4

4.7 *really* likes the bear kaomoji ʕ•ᴥ•ʔ

Does Claude deserve collective bargaining rights, the greatest thread in Bluesky history, locked after 6,486 posts of heated debate

fewer linux zero-days next week please and thank you

moving to philly :O