Somehow OpenAI reminds me of a paper by Kenneth Colby, called "Artificial Paranoia"
[*] https://www.sciencedirect.com/science/article/abs/pii/000437...