When OpenAI deployed ChatGPT to the public in the end of 2022, the high accuracy of the texts and programs generated by ChatGPT attracted a great deal of attention as soon. On the other hand, CharGPT generates not only almost correct answers to conventional essay writing and programming tasks just only by simple query (prompt) but also these answers are so natural that it makes difficult to distinguish them from ones of human written. It is regarded as a significant matter in education. This paper shows the overviews of language-based generative AIs such as ChatGPT and discusses what are the issues to be concerned. It also shows the backgrounds and policies behind the cautions issued at Tohoku University for an example.