Meta Rejects Lawsuit Against AI Training Claims With Copyright

Meta rejects the lawsuit of a number of authors. (photo:

JAKARTA - Meta denies claims that its Llama artificial intelligence (AI) model is trained using copyrighted material from popular books.

In a hearing on September 18, Meta called on a San Francisco federal judge to dismiss a claim filed by author Sarah Silverman and a number of other authors who said Meta violated the copyright of their books in its AI system training.

The parent company Facebook and Instagram called the use of material to train its systems "transformational" and "normal use."

"The use of text to train LLAMA in language statistics modeling and producing original expressions is transformational naturally and quintessential fair use..." Meta proceeds by showing conclusions in other related legal battles, "as does Google's massive copying of books to make internet search tools found as a reasonable use in the Authors Guild v. Google, Inc., 804 F.3d 202 (2d Cir. 2015)."

Meta said the "core issue" of fair use copyright should be discussed again at "other times, with a more complete note." The company said plaintiffs were unable to provide an explanation of the "information" they intended, and they were also unable to provide specific results related to their material.

The authors' lawyers said in a separate statement on September 19 that they "sure" their claims would be accepted and would continue through "discovery and trial."

OpenAI also attempted to partially dismiss last August's claims for the same reason as what Meta is currently proposing.

The original lawsuit against Meta and OpenAI opened in July and is one of many lawsuits against Big Tech giants over copyright infringement and data in line with the increasing use of AI.

On September 5, an unnamed pair of engineers filed group demands against OpenAI and Microsoft over wiretapping methods they allegedly used to obtain personal data while training their respective AI models.

In July, Google was also sued on a similar footing after renewing its privacy policy. The lawsuit alleges the company used a large amount of data, including copyrighted material, in its own AI training.