I'm not sure what they are afraid of. Is GPT-5 going to write write a meth recipe in a pirate voice, or something? Seems like a big red herring. Releasing AI tech to regular people is a good thing. What we should really be worrying about is government AI. We're already seeing some scary surveillance systems, and it won't be long until autonomous AIs are armed and trained to kill. Openai isn't the problem. How about we 'pause' the evil dystopian stuff instead?