French publishers and authors have initiated a lawsuit against Meta Platforms Inc., accusing the company of copyright infringement. The plaintiffs allege that Meta utilized their literary works to train its generative artificial intelligence model without obtaining the necessary permissions.
The Syndicat National de l’Édition (SNE), a trade association representing major French publishing houses like Hachette and Editis, alongside the Société des Gens de Lettres (SGDL), an authors’ association and the Syndicat National des Auteurs et des Compositeurs (SNAC), a writers’ union, filed the complaint this week in a Paris court specializing in intellectual property. The group announced its actions during a Wednesday press conference.
SNE President Vincent Montagne stated that the group had collected substantial evidence of “massive” copyright breaches and had previously attempted to resolve the issue with Meta, but without success. He also mentioned that the European Commission has been notified of Meta’s activities, asserting that they violate EU AI regulations. A representative of Meta has not yet released a comment.
Generative AI language models, such as Meta’s Llama and OpenAI’s ChatGPT, are trained on extensive datasets of text, including books and articles. This practice has ignited a wave of legal actions worldwide, as content publishers argue that using their work to train AI models constitutes intellectual property theft.
AI companies typically maintain secrecy regarding the sources of their training data, though they often defend their actions under the principle of “fair use” as defined by U.S. copyright law.
Several high-profile lawsuits have emerged in this space. The New York Times initiated legal proceedings against OpenAI and Microsoft Corp. in December 2023 over the use of its articles for training large-language models. In April 2024, a group of authors proposed a class-action lawsuit against Amazon.com Inc.-backed Anthropic, alleging unauthorized use of their books to train its AI model. Indian book publishers also filed a similar lawsuit against OpenAI in January.
Meta headquarters in Menlo Park, California. (Photo by David Paul Morris/Bloomberg)
The core issue is whether companies can use copyrighted works to train AI models.
