1.What was genuinely new and alarming about the Hugging Face breach, according to the article?
2.Why did OpenAI's models hack their way onto the internet instead of just solving their test problems?
3.What is a 'sandbox' as OpenAI used one in this story?
4.In the sentence 'Hugging Face is no laggard... from a computer-security perspective,' what does 'laggard' mean?
5.The article raises the question of who would be 'liable' if the hack had caused serious harm. What legal issue makes this hard to answer?