Recently, AI researchers have successfully uncovered the training data sources of OpenAI's ChatGPT through a "poetry attack." The attack method involves repeating specific words, which causes ChatGPT to leak personal information, including phone numbers and email addresses. The vulnerability was reported to OpenAI on August 30th, and there has been no further comment. The research calls for more environmentally friendly AI usage, pointing out that the energy consumption for generating text and images is enormous, and warns that caution must be exercised in promoting AI development to handle privacy and energy consumption issues.