Meta is required to respond to allegations of deliberately stripping copyright management information (CMI) from materials used to train its AI models, following a ruling by the court in a class-action lawsuit filed by a group of writers.
Legal proceedings against the company have been ongoing since July 2023, with plaintiffs asserting that Meta unlawfully utilized their works to train its neural networks. In January 2025, the writers introduced new accusations, claiming that the company knowingly trained its models on materials containing CMI—information regarding authorship, licensing, usage terms, and the date of creation—only to intentionally remove such metadata.
The court ruled that the lawsuit, alleging violations of the U.S. Digital Millennium Copyright Act (DMCA), may proceed, as the accusations provide “plausible grounds” to believe that Meta erased CMI to obscure its use of copyrighted content. This decision increases the likelihood that the case will proceed to trial or be resolved through a settlement.
Meta has already acknowledged that it used the Books3 dataset to train Llama 1, which was found to include copyrighted works. However, the company successfully had a separate claim—related to violations of California’s Computer Data Access and Fraud Act (CDAFA)—dismissed.
Legal experts noted that the ruling does not establish a violation of the fair use doctrine but affirms that the plaintiffs’ attorneys have substantiated their claims with concrete evidence. The judge also emphasized that the alleged DMCA violations may be reconsidered at the final ruling stage. As of now, Meta has not issued an official statement regarding the court’s decision.
