3 ms·
I tried to search around for this, but are any organisations doing serious work around detecting ML generated content? Like a (comparatively lame) real life bla
by fhd2 4y ago
I tried to search around for this, but are any organisations doing serious work around detecting ML generated content? Like a (comparatively lame) real life blade runner. Seems there's a lot of stuff around adversarial attacks, but that's not quite what I mean.
- nojs 4y agoYou might be interested in this post: https://scottaaronson.blog/?p=6823 https://scottaaronson.blog/?p=6823 Search for “My main project so far has been a tool for statistically watermarking…”
- mijoharas 4y agoI heard about a startup doing exactly this yesterday evening (pen testing AI models for adversarial attacks). I think they're mostly targeting images and convnets at the moment. I can't recall the name of the company at the moment, but will update here with a reply when I find it.