I wonder if OpenAI's goal is to keep potentially dangerous tech from leaking while making their supposedly safe solution so cheap that it discourages reckless competition.
They have stated in their interviews that developing an AGI could be great or terrible for humanity, and it's ultimately uncontrollable, so the best way to get the best outcome is to set the initial conditions under which the technology gets developed to be as favorable as possible.
I wonder if an AI researcher would love to build a better interpreter for reverse engineering LLMs and would want to try it on an LLM clone of OpenAI's setup.
The cloud-service moat is the strongest security for OpenAI.