i have determined that we are absolutely fucked but feel free to wait for a lawyer to confirmhttps://twitter.com/hipsterelectron/status/1410391147868946433?s=21 …
-
-
Antwort an @hipsterelectron @iwasleeg und
you're allowing the “this was mediated by tech” logic confuse you: if a snippet is big enough and you have copyright on a similar enough bit of code -- the mechanism by which someone copied it doesn't matter copyright protects the expression of an idea regardless of copy method
2 Antworten 0 Retweets 5 Gefällt mir -
Antwort an @luisbruno @hipsterelectron und
if a large snippet is constructed by you+copilot in multiple steps and ends up looking just like someone else's, you done goofed ;) i would also like to attach to the model itself, though -- a ML model is a derivative work on its own, and i which this wasn't still being debated.
2 Antworten 0 Retweets 5 Gefällt mir -
Antwort an @luisbruno @hipsterelectron und
Yeah, that latter bit was my original point. It's easy to hide a derivative work generated by the model, but the model itself MUST comply with the licenses it is trying to launder.
1 Antwort 1 Retweet 7 Gefällt mir -
Antwort an @NoraDotCodes @hipsterelectron und
my apologies, i noticed it earlier but ran out of characters to fully word it; yes you had mentioned it, and word it better than i did ;)
2 Antworten 0 Retweets 3 Gefällt mir -
Antwort an @luisbruno @NoraDotCodes und
oh wow. now that you mention it, it's a very interesting question whether it's even possible for an ML model to comply with the GPL.
1 Antwort 1 Retweet 8 Gefällt mir -
Antwort an @ireneista @luisbruno und
the GPL goes into considerable detail on what it means to provide the "source" of something.
2 Antworten 2 Retweets 7 Gefällt mir -
Antwort an @ireneista @luisbruno und
Yeah. They would have to provide the scripts they used to train the model, and all the training material...
1 Antwort 1 Retweet 5 Gefällt mir -
Antwort an @NoraDotCodes @ireneista und
sorry, i don't see it: that's the AGPL territory, no? for GPL, providing the original training data without modification should meet section #1 of the GPL
1 Antwort 0 Retweets 3 Gefällt mir -
Antwort an @luisbruno @NoraDotCodes und
yeah, you're correct. we went back and re-read the GPL just now to look for the relevant wording and that's when we were reminded of the AGPL.
2 Antworten 0 Retweets 7 Gefällt mir
Are we sure the training set didn't include AGPL code, though?
-
-
Antwort an @NoraDotCodes @luisbruno und
it would be pretty difficult for GitHub to filter things out that way, given that a single repo may combine code under different licenses. but we shouldn't have to speculate - it's a reasonable question that GitHub should address.
1 Antwort 0 Retweets 3 Gefällt mir -
Antwort an @ireneista @NoraDotCodes und
if they're not willing to address it, that's damning in its own right.
1 Antwort 1 Retweet 5 Gefällt mir - Antworten anzeigen
Neue Unterhaltung -
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.