Original link: OpenAI declares AI race “over” if training on copyrighted works isn’t fair use / ArsTechnica.
I think they are on thin ice.
If they can prove that AI does not reproduce material it is trained on then AI are not violating copyright. But I don't think they can, because it will happen that AI recreate large parts of copyrighted material.
And the argument that it is OK to violate copyright to keep US ahead of China in AI race does not seem like a valid legal argument to me.