One reason we may be having difficulty with this section is that "data" can be literally anything from basic facts, which is what lay people think of as "data" to professional works of fiction or visual art, which lay people think of as more than mere data. And those lay intuitions are legally relevant.

As a result of these distinctions, and the many other rights issues that may arise with certain types of data - e.g., health information or nude pictures of real people, this area seems like the most ethically complicated.

I don't have a good answer here, I'm just trying to call out why this feels more challenging. Maybe that will help shake some ideas loose as we continue to iterate.

It might be helpful to include a couple of examples of what we are talking about here. I think this is referring to many of the ethical questions around impact on jobs or specific uses of AI to manipulate or otherwise harm people. And are we trying to say here that by creating a definition of "open" for the purposes of modification, we are not weighing in on those ethical questions?