Researchers Poke Holes in Safety Controls of ChatGPT and Other Chatbots (New York Times)