[slashtab]

Meanwhile nothing for mega corporation pirating data to train AI.

[ProofHouse]

It’s not pirating. It’s transformative and fair use. Derivative even in some cases. Each piece of content is but a grain of sand on an island.

It’s called the open internet.

[ruthmarx]

It's pirating when these scrapers are ignoring terms of user and whatever the AI equivalent to robots.txt is as has been widely reported as happening.

[bredren]

Key here would be transformation rather than reproduction?

Youtube is mentioned in the 2013 brief:

>b. According to the YouTube “Terms of Service,” users who upload content to YouTube retain all of their ownership rights in their content. By uploading their content to YouTube, however, such users grant YouTube a license to use, reproduce, and distribute such content. >

>c. In general, the further reproduction and distribution of videos that are taken from the Youtube.com platform violates the copyright of the individual who uploaded that video to Youtube.com.

[chgs]

I my country making a copy of copyrighted works is illegal unless you have permission from the copyright holder.

It’s impossible to use a copyrighted work to train without making a copy of it.

[hnisoss]

So how much of a book of poems can I assemble from other books of poems and spit out as an ebook on amazon before it is not "transformative and fair use"? Few words, sentence, chapter?

[suprjami]

It's called the open internet when the same rule applies to everybody.

This is one rule for corporations and another rule for individuals.

Additionally, there is often with little or no recourse when the latter are falsely accused.

Look and the frustration of YouTubers evoking DMCA takedowns for including 5 seconds of a commercial song accidentally.

[trueismywork]

Training is not pirating, but generating copyrighted data is.

[_giorgio_]

You pirated all the books you have read, all the movies you have watched, and you're pirating this message right now.

Show me the money.