Meta Says Its AI Models Don’t Infringe Authors’ Copyrights

In an attempt to dismiss a high-profile copyright infringement class action, Meta Platforms Inc. said its Large Language Model Meta AI doesn’t infringe a group of authors’ copyrights because its outputs aren’t substantially similar to the authors’ works.

Sarah Silverman, along with two other authors, sued Meta and OpenAI in July. The authors alleged the companies improperly used their works as training material without their permission for generative AI software programs and removed their copyright management information.

Since copyright law only protects how information is expressed and the authors aren’t arguing LLaMA and the works are similar, the authors failed ...