My company doesn’t train models. We just use them. Can we still face copyright risk?
This is a very common blind spot, and that is often overlooked.
People assume that the liability about copyright rests on AI developers, not on the people that use it. But you must keep in mind that you create copies when you upload, paste, or connect third party works to an AI tool. That triggers the reproduction rights.
Also, at the output level, the output may be reproducing a substantial part of a work, and that too may trigger copyright.
It might also be a derivative of a work that the AI was trained on or prompted with, and that too may generate an infringing liability.
So, remember: even though you’re not the model developer, as a user of the model, you can also trigger copyright liability.