Conversation

Notices

1. Embed this notice
   Peter Henderson (phend@mastodon.social)'s status on Monday, 16-Jan-2023 01:20:52 JST Peter Henderson

   New lawsuit against stable diffusion models and associated companies. Unlike the copilot lawsuit there is an argument for direct infringement here arguing that the model itself is essentially a compressed database of copyrighted images.

   https://stablediffusionlitigation.com/pdf/00201/1-1-stable-diffusion-complaint.pdf

   • Embed this notice
     Alexandre Oliva (lxo@gnusocial.net)'s status on Monday, 16-Jan-2023 01:20:45 JST Alexandre Oliva

     in reply to

     • Iron Bug
     but the drawing of the same picture from memory would be a derivative work, no?
   • Embed this notice
     Iron Bug (iron_bug@friendica.ironbug.org)'s status on Monday, 16-Jan-2023 01:20:47 JST Iron Bug

     in reply to
     @phend it's like to say that any person that has seen a copyrighted image can draw the same picture. but this is not a digital copy of any kind, by definition.
   • Embed this notice
     翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Monday, 16-Jan-2023 19:17:38 JST 翠星石

     in reply to

     • Iron Bug
     @iron_bug >actually trained on publicly accessible data that are barely copyrighted.
     What are you even on about?
     
     Works are either copyrightable and automatically copyrighted (Berne convention), or not copyrighted.
     
     Anything that counts as a creative work is copyrightable.
     
     Facts like a list of ordered words in a dictionary or a list of street names aren't copyrightable, although if you were to creatively format this information into a book, the whole book qualifies for copyright.
     
     
     Most models are trained on a large amount of publicly available copyrighted works, alongside a small amount of purely factual information.
     
     Just because a copyrighted work is publicly available doesn't make it not copyrighted.
   • Embed this notice
     Iron Bug (iron_bug@friendica.ironbug.org)'s status on Monday, 16-Jan-2023 19:17:39 JST Iron Bug

     in reply to
     @phend why do they consider that a model was trained on copyrighted data? the most models in research are actually trained on publicly accessible data that are barely copyrighted.