$45k is "new car" money, plenty of individuals on this site could afford it if they wanted. Don't even have to be that serious about it.

Exactly, I can't imagine that the OpenAI team would be this naive. Their strategy will do precious little to stop nefarious actors who mean business. It will just narrow the circle of well meaning people who could've participated in research.

Besides, it's only a question of time before someone publishes the full model.

The whole gatekeeping thing just feels like empty posturing.

It's not quite so simple: AFAICS they have not open-sourced the training code, only one for sampling from the pre-trained model.

what would they do with it? It is a text decorator, it can't make sense of itself. maybe they can make 1000 fake facebook accounts that post a lot. still, i doubt facebook uses text to detect bots, its easier to use ip/network patterns.