[spacecadet]

I had a client train an AI on images I created without extended usage and so added adversarial noise to the images next time around. The models I tested with misclassified the images and image generation seemed broken, so Im curious how it impacted their attempts, if they even attempted it again, I don't know. I don't expect them to come to me and ask why their model is so interested in ducks...

[constantcrying]

Companies put measures into place to make sure that their software only functions correctly with a paying customer. Some Video Games include intentional problem when the game believes IP laws are violated.

[pixl97]

If the client is sophisticated enough to realize that I don't believe it's difficult to prefilter to remove said noise.

[spacecadet]

For sure. It was a test to see what would happen. I heard nothing, but also didn't see my work appear again- who knows...