As for the online limiting, it's as simple as CYA.

There are multiple stable diffusion installs you can do on your own [1] and run whatever wild queries you want.

[1] https://github.com/invoke-ai/InvokeAI