Is it ethical to expect people who uploaded an image to the Internet a decade ago—or had it uploaded illegally for them—to accept that their data is now being used to train the future of AI without recourse?