My question is somewhat similar to this one, but I am interested in a case where there is no explicit allowance for models that are unable to reconstruct their training data.

To be concrete let's focus on training data licensed with creative commons v4 (BY, BY-NC and BY-ND.)

Suppose I create a machine-learned model by training it on tens of thousands of materials made available by these CC licenses. I then use the model and its outputs for commercial purposes.

  1. Have I made use of the training material in the BY sense?
  2. Have I made commercial use of the material in the BY-NC sense?
  3. Are the model or its outputs derivative works in the BY-ND sense?

It is not possible to recreate any of the training materials from the model or its outputs. A typical person familiar with the training materials would not be reminded of any specific one by the model or its outputs.

EDIT: As a rider that only occurred to me afterwards, would the model and its outputs fall under fair use?

2 Answers 2

  1. Under BY, you have a duty to provide attribution for your adaptive use of their product. It can be simple: "Made possible through the use of..." There are no changes to the data, as they are not even presented. So no further qualifying remarks needed.

  2. Under BY-NC provisions, you have no obligation in excess of that covered in BY (above). The materials are no longer present in your product. So no direct relationship there. And any attempt to tie those materials directly to the product or its output would be comical. I take it that you could not use your own product to either output anything recognisable as their property or create any, even vaguely, proximal relationship between the tools and the output.

  3. While model output is derivative of it's input, you don't even come close to reaching the standards in BY-ND. So a hammer was essential in building a machine that makes nail guns. Right.

Fair use. Where did you get the idea that fair use can come into focus here at all? Sorry, mate. That one's a non-starter.


The links you supply seem to give the answer: They stipulate that you must give credit to the materials you used to train your model.

I'm not a trained legal professional. I am just reading back to you what you supplied to me in the links.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .