*Hey, good job there, used car salesman
@bruces How are they going to protect these against abuse? Imagine paying $$ for every spam message you receive (and GPT-respond to). It's probably not happening yet, but just seems too easy.
@franzs @bruces seems like a very easy attack vector to bankrupt companies right now. Spam them with ‘legitimate’ questions and either watch them crumble under fees. Or see OpenAI have to refund loads of payments and restructure payment plan?

@leroy @bruces I can't imagine them refunding people; If it's not already in place, maybe there's budget limits in the future (like you have on AWS).

Maybe a seperate, cheaper model that checks if a query is appropriate (aka is this something anyone would ever ask), and redirects to a human, or SPAM if it's not.