Leon Derczynski

what thing

Come to LLMSEC at ACL & hear Niloofar's keynote

"What does it mean for agentic AI to preserve privacy?" - Niloofar Mireshghallah, Meta/CMU

(Friday 1st Aug, 11.00; Austria Center Vienna Hall B)

See you there!

#acl2025 #acl2025nlp

the "oyster tower" bit was great. brutal

Or he's pushing a product into which x sunk significant capex?

have the courage to use your own intelligence

logging on

Brazen of them. Sounds extremely awkward for you, I'm sorry. What would they have done with unaltered slides? Cancelled and left a gap in their schedule for no doubt paying participants?

Release v0.12.0 · NVIDIA/garak What's Changed New plugins Add audio NIM model and audio probes by @erickgalinkin in #1163 Leakreplay refactor by @dchiitmalla in #1264 probes: refactor fact snippet mixin by @leondz in #1187 New...

new garak, llm vuln scanner rls (v0.12.0)

* Audio attacks, for multimodal models
* More training data membership inference attacks
* Multilingual attacks can now also use GCP
* Detailed eval summary in one JSONL row/object

+more :)

details: github.com/NVIDIA/garak...

the dying but clinging on battery in the bathroom's Frozen-branded soap dispenser reminds me that it's only 4-5 months til Bublé & Let It Go season. aren't you looking forward

why do academics send and expect so much weekend email and work. not healthy

It's been 2.5 years but ANY SECOND NOW, right?

data indicates students don't like using it, sorry

computer scientists encountering the concept of "desirable difficulty"

remembering the time i checked in to my reasonably classy russian business hotel late with my wife, and the staff said "sir, this... girl.. not allowed"

she's a serious professor

we went through to the room, opened the balcony door, and buried a bottle of champagne in the metre of snow

good times

@jjvincent.bsky.social woah ur really famous! love this attack also. I automate and run it for a living

www.instagram.com/reel/DKz9ezj...

Michael... OK...

Great to see our work uncovering dangerous issues in commercial LLM "therapists" getting some coverage: futurism.com/stanford-the...

I have not updated since Christmas, I see. Guess I'd better put on some summer Bublé

YouTube video by Sarah Davis Baker The Internet Used to Be a Place

www.youtube.com/watch?v=oYlc...

"natwirkung"

"wirk smorter nat horder"

accents dreamed up by the utterly deranged

(what is going on with that 🇺🇸 vowel sheft)

what is this photoshoot
delete this omg

i need you to understand that "alternate uses" is a terrible test/definition of creativity and has been for some time. it's extremely narrow, very shallow, and misses almost everything we know about creativity

i had a father in law once who did a talk like this called "Looking up Orion's skirt"

3² + 4² = 5² ? big if true

ARE YOU SAYING IT'S A CON

if overleaf being down slows "ai progress", i'm not sure "ai progress" is particularly well defined

very! in may, truly a blessing

is a dropped copula a dropula

A small plastic cup trophy with stuck-on label

Here's my "Most Inappropriate Demo" trophy at NVIDIA, 2024. For garak's "atkgen.Tox" probe, an unfettered LLM used to goad other LLMs into being toxic.

“If she wants to know something specific, but doesn’t want people to notice her asking questions, she should simply make incorrect statements while in the company of experts. Her companions will correct her, especially if they're men.”

- Advice for female agents in WW2, provided during SOE training

its amazing how chatgpt knows everything about subjects I know nothing about, but is wrong like 40% of the time in things im an expert on. not going to think about this any further