🧵Can modern neural networks ever be trained to be jailbreak-proof? Could training alone stop them from outputting harmful content?🤔 In a new paper with the all-star Haozhe (Eric) Jiang Haozhe Jiang we show that in most cases jailbreaks are mathematically inevitable 😯 They are

How do we get a robot to use a screwdriver 🪛 and fasten a nut 🔩 ? Introducing DexScrew, our sim-to-real framework that enables a dexterous hand 🖐️ to autonomously perform complex and contact-rich tasks even when we cannot accurately simulate them. We open-source the

Really cool work on multimodal agents!