Meta downloaded an explicitly pirated data set of millions of published works on which to train their AI tool. The Atlantic has helpfully made that data set searchable. It appears to include 4 of my plays 👍🏻 www.theatlantic.com/technology/a... - ThreadSky

kieranhurley.bsky.social • 35 days ago

Meta downloaded an explicitly pirated data set of millions of published works on which to train their AI tool. The Atlantic has helpfully made that data set searchable. It appears to include 4 of my plays 👍🏻

https://www.theatlantic.com/technology/archive/2025/03/libgen-meta-openai/682093/

1 / 2

Comments

peterfurmedge.bsky.social•35 days ago

My small contribution to Zuckerberg's accumulation of everyone else's work.

kieranhurley.bsky.social•35 days ago

Worth saying that for from being surprising, this is now beyond widespread, and something our current government is actively seeking to further entrench by changing copyright law to further facilitate this mass theft by Zuckerberg and his ilk, at the expense of the rest of us

davidpattie.bsky.social•35 days ago

Completely agree (& yeah. My stuff's been pirated too).

cdelgadogarcia.bsky.social•35 days ago

Oh wow, mine too...

tomusher.bsky.social•35 days ago

is there any way you can do something about this legally? surely this actually comes under some kind of copyright law at present at the very least?

kieranhurley.bsky.social•35 days ago

I mean, it's probably not entirely healthy that my first reaction is to chuckle and impotently shrug it off, but I do think the fact of it being so totally widespread sort of bludgeons it into being basically inactionable by individuals idk

tomusher.bsky.social•35 days ago

i really dont think that's the case, they are literally pirating your work, or using pirated versions of your work. that's definitely actionable surely

kieranhurley.bsky.social•35 days ago

Their legal argument hinges on an interpretation of "fair use" in copyright law (bollocks but working from them). Also in the article it says something about how it's impossible to prove which parts of the dataset were used to train their AI and which weren't, and I guess that's true.

kieranhurley.bsky.social•35 days ago

Agree it's clearly obvious they've ripped millions of writers off! It's very shady and frustrating!

csvich.bsky.social•35 days ago

we make two cents to begin with. argh. this is madness

msmaev.bsky.social•35 days ago

I checked, and while my play did not appear, they appear to have pirated a bunch of my father’s Irish language newspaper articles, which begs several questions.

duncangates.bsky.social•35 days ago

yet another longlist i fail to bloody make

stephenhodge.bsky.social•35 days ago

i go well out of my way to side-step meta in its multiple incarnations. not sure whether to laugh or cry about how my small pirated input might détourné any subsequent output.

Comments

Posting Rules

Reply