…year so far: “radioactive data.” They write:

In the subfield of computer vision, researchers at Meta have demonstrated that images produced by AI models can be identified as AI- generated if they are trained on “radioactive data”—that is, images that have been imperceptibly altered to slightly distort the training process. This detection is possible even when as little as 1% of a model’s training data is radioactive and even when the visual outputs of the model look virtually identical to normal images. It may be possible to build language models that produce more detectable outputs by similarly training them on radioactive data; however, this possibility has not been extensively explored, and the approach may ultimately not work.

No one is sure exactly how (or if) this would work; it’s much easier to alter an image imperceptibly than it is text. But the basic idea would be to “require proliferators to engage in secretive posting of large amounts of content online,” they write, in hopes that models trained on it would produce text that could be traced back to those “radioactive” posts. If by now you’re thinking “that’s bonkers,” you’re not alone. Among other things, the authors note, this nuke-the-web plan “raises strong ethical concerns regarding the authority of any government or company to deliberately reshape the internet so drastically.” And even if someone did go to those lengths, they write, “it is unclear whether this retraining would result in more detectable outputs, and thus detectable influence operations.”

Regarding the “ethical concerns”, I have to say: you can’t cheat an honest man. If this altered material is posted publicly but without permission for use, and it happens to be slurped down by those with no intention to license it, should I feel bad for them? The republication of scraped publicly posted content is already a travesty before neural networks are brought in to launder the IP concerns. Even I on this proudly two-bit, proudly dog-and-pony website have had my writing stolen1!

Anyway, I think whatever mischief folks can get up to that gives the creators of these things a hard time is probably good on balance – even if just to rattle the bars of the narrative that it’s beneficent “research”.

  1. I sent one (1) email to the effect of “I see what you did there” and the stolen piece was deleted, but it leaves one wondering if there are more I never found…