x
OpenAI’s surveillance language has many potential loopholes an — LessWrong