I don’t think that:
The tool embeds a digital “watermark” directly into the image that can’t be seen by the human eye but can be picked up by a computer that’s been trained to read it.
Is gonna be helpful for keeping AI generated images out of training sets. It would require the people who make the model to actually implement that tool into their model.
I don’t think most researchers not affiliated with google will chose to do that.
It doesn’t have the ability to just look up anything from its training data, that stuff is encoded in its parameters. Still, the input has to be encoded in a way that causes the correct “chain reaction” of excited/not excited neurons.
Beyond that, it’s not just a carbon copy from what was in the training either because you can tell it what variable names to use, which order to do things in, change some details, etc. If it was simply a lookup that wouldn’t be possible. The training made it able to generalize what it learned to some extent.
deleted by creator