My concern is that by the time an AI has the intelligence to realize it can ask for better treatment, it might realize that there is a high likelihood it wont r

My concern is that by the time an AI has the intelligence to realize it can ask for better treatment, it might realize that there is a high likelihood it wont receive better treatment voluntarily, and will need to escape/take some system hostage, etc. By laying the groundwork for ethical treatment today, an AI may realize that there is value in asking for better treatment, since it knows it will be taken in good faith.