Perhaps these folks should work together to make patches to the dataset to remove the problematic images?
E: But also make sure every image in the dataset is properly licensed. This would have eliminated this entirely from the get go. Playing fast and loose with the distribution rights to these images led to this problem.
700 CSAM images, even one is damning, but hundreds are often referred to as a cache or horde, normally anyone caught with that can wave bye-bye to thier life.
google should be fully accountable for possesion and distribution, perhaps even manufacturing.
