oh my gods. they literally have no shame about this. GitHub Support just straight up confirmed in an email that yes, they used all public GitHub code, for Codex/Copilot regardless of license.pic.twitter.com/pFTqbvnTEK
Diesen Thread anzeigen
-
Is there any precedent for training an AI on copyrighted content being a violation of said copyright?
My take: training AI on publicly available code is perfectly legal as far as I can tell, but if your AI is so underconstrained that it starts spitting out verbatim snippets of code, and you serve them to other people without the appropriate license, that's copyright infringement.
-
-
I.e. that's not an AI, that's a shitty storage system.
-
until someone calls it "convergent synthesis"
- Antworten anzeigen
Neue Unterhaltung -
-
-
I think the issue here is that with a sufficiently complex neural network, there’s no way to prevent overfitting / verbatim copying of text since the structures used are so opaque
-
Exactly. Which is why the lawyers should never have approved this. It's GitHub's responsibility to not infringe copyright. "We want to sell a thing that accidentally violates copyright sometimes" doesn't fly.
- Antworten anzeigen
Neue Unterhaltung -
-
-
This is also assuming you don't copy/convey to others the model itself, which is a derivative work of everything it was trained on.
Danke! Twitter wird dies nutzen, um deine Timeline zu verbessern. Rückgängig machenRückgängig machen
-
Das Laden scheint etwas zu dauern.
Twitter ist möglicherweise überlastet oder hat einen vorübergehenden Schluckauf. Probiere es erneut oder besuche Twitter Status für weitere Informationen.