If you were the CEO of OpenAI, and you knew someday there would be a need to verify if someone is human, what would your plan be?
Non-humans could potentially have a public key; I'm a fan of switching to public keys in general but it doesn't solve the problem of assuring "unique humanness".
Have I met them? Check. Has someone I know and trust met them? Check. Has the government met them? Half check. Has the google met them? Quarter check.