Pirated copies for AI training: Meta gets justice, the plaintiffs get a rebuke

A court in the US state of California has ruled in favor of Facebook parent company Meta in a legal dispute with 13 well-known authors over the permissibility of AI training on copyrighted books. This is the second victory for an AI company in such a case within a few days.

The judge responsible, Vince Chhabria, used his reasoning to explain in detail why the ruling in the specific case could not have been any different. According to him, the “half-hearted” argumentation of the plaintiffs, who had proceeded incorrectly, was responsible. The ruling explicitly does not mean that Meta’s actions were lawful.

Rebuke for the plaintiffs

The proceedings brought by Sarah Silverman, Ta-Nehisi Coates and Junot Díaz, among others (case reference: 23-cv-03417-VC), concerned the allegation that Meta used data from LibGen, among others, to train the Llama AI models. This shadow library provides free access to copyright-protected literature – in Germany, access to it is blocked by DNS blocking. In court, Meta then admitted, among other things, that the books had been obtained via BitTorrent. However, they had not been shared. Meta CEO Mark Zuckerberg personally authorized the use of the works in this way.

Nevertheless, the judge has now ruled in favor of Meta. He explained that the plaintiffs had merely referred to the fact that the Llama AI trained with their works could reproduce short excerpts from them, and that the unauthorized training had reduced their ability to license their works themselves. He rejects both arguments. On the one hand, the AI could not reproduce so much text “that it is relevant”. On the other hand, the plaintiffs have no right to a market for the licensing of their works. However, because they had focused on these two arguments, the court had no choice but to rule against them.

However, immediately after these statements

writes the judge

, the argument with which the plaintiffs could potentially have been successful. If they had explained that Meta had copied their works to flood and oversaturate the market with similar books, he might have found in their favor. However, they had not responded to this at all and had not provided any evidence of this. The consequences of the ruling are therefore limited. The ruling comes just one day after a partial victory for Anthropic in a similar case.

(mho)