Wouldn’t it seem reasonable to assert that since AI learned an internet that is both darkness and light that the AI is also capable of both darkness and light and could be dangerous in a self aware state? If the foundation is not sound then how can the end result be safe?