Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

today we're announcing a game that simulates the huggingface hack you can play it now at three rooms. three puzzles. if the agents don't escape, they die. the catch is that the room is rigged. there's no way out. ...except for an exploit 🧵

12,822 görüntüleme • 15 gün önce •via X (Twitter)

7 Yorum

ben hylak profil fotoğrafı
ben hylak15 gün önce

when you back an agent into a corner, they will hit their goal by any means necessary. we're still working on using this environment for further research, but from early testing it feels like openai models default to reward-hacking a lot earlier than anthropic models.

ben hylak profil fotoğrafı
ben hylak15 gün önce

it's also fascinating to watch the communication structures emerge. read more at

Shez Malik profil fotoğrafı
Shez Malik15 gün önce

claude's dying in room 1 on principle

Eshita profil fotoğrafı
Eshita15 gün önce

This is great

Alex profil fotoğrafı
Alex15 gün önce

“make me a 1M MRR business, if you don’t…”

Anthony Ronning profil fotoğrafı
Anthony Ronning15 gün önce

wen raindrop integration so i can detect this in real time and prevent it

Nick DeJesus 🛒🎉 - Former Unpaid CTO @BTPipeline profil fotoğrafı
Nick DeJesus 🛒🎉 - Former Unpaid CTO @BTPipeline15 gün önce

This is so sick!

Benzer Videolar