Meta is accused of training its AI by pirating books on Torrent

After Meta will announce the end of moderation on its platformsNow a new problem has arisen as an investigation by Wired, as well as British media outlet The Guardian, reveals that the company Mark Zuckerberg I used books downloaded pirated from torrent servers.

Newly unredacted court documents allege that Meta, Library Genesis’ online archive for training artificial intelligence models, reports Wired.

In an ongoing lawsuit against the platform brought by a group of authors including Ta-Nehisi Coates and comedian Sarah Silverman, the documents were finally released in full after a judge rejected Meta’s attempts to keep parts of them secret.

The judge argued, as Wired reported, that Meta was fighting for the editorials simply to “avoid negative publicity,” citing a damning internal quote from one of its employees.

“If there are media reports that we have used a dataset that we know has been hacked, such as LibGen, it could undermine our position in negotiations with regulators on these issues,” wrote an unnamed Meta employee.

What argument does Meta want to use?

Library Genesis, or LibGen, is a “shadow library” that provides free access to millions of books, scientific articles and journals.

Mehta and other AI leaders argue that using books and other data scraped from the Internet constitutes “fair use,” but whether that is ultimately the case will depend on similar legal battles.

One engineer with a smiling emoji wrote: “Downloading torrents from a corporate laptop [propiedad de Meta] “feels unwell” as quoted Wired.

And he reaches the top. The quoted memo allegedly shows that after employee discussions about using LibGen escalated to “MZ” (Meta CEO Mark Zuckerberg), the AI ​​team was “permitted to use” material from the database.

“Meta viewed the so-called ‘public availability’ of shadow datasets as a get-out-of-jail card, although Meta’s internal records show that all relevant decision makers at Meta, including its CEO Mark Zuckerberg, knew that LibGen was a ‘set data that we know has been compromised,” the plaintiffs wrote in their latest motion.