As if the models have not seen the open source software before. That should be considered in the upcoming ruling. Technically the models are trained on exactly that.