Conversation

Notices

1. Ars Technica (arstechnica@mastodon.social)'s status on Tuesday, 06-Feb-2024 06:06:28 JST Ars Technica

   4chan daily challenge sparked deluge of explicit AI Taylor Swift images

   New AI image filters are "just another obstacle to ‘defeat,’" researcher says.

   https://arstechnica.com/tech-policy/2024/02/4chan-daily-challenge-sparked-deluge-of-explicit-ai-taylor-swift-images/?utm_brand=arstechnica&utm_social-type=owned&utm_source=mastodon&utm_medium=social

   • Howard Chu @ Symas (hyc@mastodon.social)'s status on Tuesday, 06-Feb-2024 06:22:30 JST Howard Chu @ Symas

     in reply to

     @arstechnica forget it, just assume all porn videos are fake and they become irrelevant. https://mastodon.social/@hyc/111064161737603346

   • zachery_delong (zachery_delong@mastodon.social)'s status on Tuesday, 06-Feb-2024 07:21:40 JST zachery_delong

     in reply to

     @arstechnica I'm very curious where this discussion goes, because I don't see any way to reliably and always remove it. Perhaps if we tried to give the model some "judgement" by parsing its own output and classifying it, but even that is going to have significant error.