It could be a lot more closed than it is, from a general public perspective. i.e. the company having this cool AI and we not even getting the opportunity to use it at all.
The human brain is a model that competes with OpenAI, and every time you look at the output, you use OpenAI outputs to develop your brain.
Thus, every user interaction violates the rules as written, and is “illegal, harmful, or abusive.”
Malicious compliance is the move. Just don’t use it!
A bit like MS Copilot for Bing / Windows which is really just rebranded ChatGPT. For the O365 version they added a lot of stuff to interact with office but their online chatbot is just the same.