Along those lines, if the model's output was actually deterministic, would the patent be bound to using the prompt with only one particular model since the output would be different, in some cases wildly different, with other models or even other versions of the same model?

My absolute wild guess as a layman is that prompts will at most get "trade secret" protection and won't be able to be copyrighted or patented.

The process is reproducible even if the outcome isn't always identical. Outside of computing and mathematics, real world processes never result in the exact same output - small variations in size, density, concentration, etc. will occur.

> if the result of the prompts are not always the same on each run?

In stabble diffusion, it seems to me that if we define a seed with the prompt, we get the same result, no?