Images generated by AI models trained on massive datasets often can’t be traced to specific training images, MIT CSAIL researchers found. Removing individual images from the dataset didn’t change outputs, complicating copyright questions.
On the one hand, sure, models generalize. It’s cool they established that.
On the other hand, the copyright argument seems to be that they stole so much data that the contribution of any individual image rounds down to zero, therefore it’s not infringing? Wha??
Are they really claiming that one infringement is a a tragedy, but a million infringements is a statistic?
On the one hand, sure, models generalize. It’s cool they established that.
On the other hand, the copyright argument seems to be that they stole so much data that the contribution of any individual image rounds down to zero, therefore it’s not infringing? Wha??
Are they really claiming that one infringement is a a tragedy, but a million infringements is a statistic?