That doesn't matter that much. If your analysis is correct then it means a (tiny) minority of OpenAI cares about AI safety. I hope this isn't the case.