Meaghan Kent, intellectual property partner at Venable LLP, explains a recent court ruling on whether the artificial intelligence training model is free to use copyrighted content.
A significant amount of litigation is underway regarding the issue of copyrighted content built into the generator AI model. Kent is Thomson Reuters v. Loss Intelligence, Claude v. In the case of Humanity and Cadley v. Metaplatform, we discuss three recent federal court decisions.
The Reuters case involved the use of AI to summarise the case principal, created by Thomson Reuters' West Publishing. The judge found that the case amounted to the establishment of a competitive product and therefore not protected by the Fair Use Act under the Copyright Act of 1976.
The second case included a group of authors who appealed to humanity to draw works to train claude.ai assistants. This case revolved around an AI model that collects data from two separate sources. A copy of existing works that are considered pirated versions of protected content, as well as certified works that have been purchased and scanned. The judge ruled that pirated work was not subject to fair use and that the principles were applied to purchases and scans. This is because these actions are likely to cause a “dilution” of the needs of the author's work in the market, and in addition, the resulting content has been fully transformed in nature to avoid copyright claims. (Editor's Note: Humanity has agreed to a $1.5 billion settlement with the author and publisher who filed the lawsuit. According to New York Times, This is the largest payment in the history of US copyright disputes. )
In the third case, including training in the Meta AI model known as the llama, the judge rejected the plaintiff's copyright infringement claims and indicated that he would have considered the “market dilution” argument, and the plaintiff presented evidence of harm.
The problems of AI and fair use are far from being solved, and Kent says that intellectual property owners and AI developers need to keep pace with the rapidly changing development. He added that there could be short-term consequences, that the copyright holder agrees to license the content to multiple AI providers.
