Подтвердите e-mail

Для публикаций, комментариев, реакций и сообщений подтвердите адрес.

Профиль

banneduser.space[DEPRECATED][REDACTED]

Профиль Vively

https://app.wafrn.net/blog/banneduser ajolotedemonico 🦎 😈he/him/it/they/e adult 32+ MDNI🔞🧬science&🎲game freak🪐 trans🏳️‍⚧️rights zone🦭VRC:@banned 🖼️ gallery try Pinksky 🦩https://pinksky.app/profile/banneduser.neocities.org Status:¡TÓXICO!☠️ ˃̶͈̀(°👅°)˂̶͈́🪐˃̶͈̀(°◡◡°)˂̶͈́

I don’t know who needs to hear this, but it’s time to take the convenient excuses to bail on Thanksgiving out of the freezer

0160

another finding in this paper is that the top AI #LLM models already have 1 in 20 odds of being exploited to generate outputs crimes, self harm, violent crimes, indiscriminate weapons like CBRN, and violate privacy at baseline. Like that’s the normal rate of that shit out of these #frontierAIs 🫧📌

Andrew Plotkin

HOLY MOTHER FORKING SHIRTBALLS IT WORKED arxiv.org/html/2511.15... blog.zarfhome.com/2023/05/sydn...

121

This and #bsky would benefit from a feature to approve followers because There are too Many Damn Bots on this shit

Catbus

i hope bsky realizes there is no more certain way of ensuring I will never follow someone back than adding a giant blue FOLLOW BACK button to my notifications

011

#zohran mamdani is going to bring #shakiraLaw to #NYC it will be REQUIRED by SOCIALIST LAW that for Hips Don’t Lie

031

Good news everyone the bots are weak to bars #poetry they #NotLikeUs

Charlie Stross

Looks like LLMs are *very* vulnerable to attack via poetic allusion: "curated poetic prompts yielded high attack-success rates (ASR), with some providers exceeding 90% ..." https://arxiv.org/html/2511.15304v1

030
Показать ещё