Two papers( Towards Understanding Jailbreak Attacks in LLMs: A Representation Space Analysis; On the generalization of training-based chatgpt detection methods) accepted to *