It's not exactly that, their rhetoric is: "This AI stuff is dangerous to world if YOU use it. But when we use it, it is good."
All entities think that they hold the key to good values, while the others are always suspected of hiding vile intentions.
Based on their actions, OpenAI seems to (rightly, in my opinion) view that some people would misuse GPT-3/etc, even if the majority would not.
Social engineering in general is effective because it's rare enough that people don't feel the need to develop policies and strategies for preventing it.
Following this logic, one would forbid schooling to lads based on the suspicion that they may "misuse" what they will learn to do some evil. One would use the alphabet to write threats to others. One would use a knife to stab others. So how do you police the right to access the knowledge? Should we restrict accessing knowledge to some few elites that some authority should decide who they are?
IMHO knowledge should be accessible to all. The accumulation of the humanity knowledge should not be controlled by an Orwellian entity. <s> Otherwise, let's start a movement demanding to wipe online universities courses from the Internet </s>