Copyright claims against AI companies have just received a possible boost.
US federal judge last week gave a brief judgment In a case brought by Tech Conglomerate Thomson Reuters against the Ross Intelligence Legal Technology Society. The judge found that the use of Ross’s content by Reuters to train the AI legal research platform violated Reuters’ intellectual property.
The result could have implications for more than 39 AI lawsuits related to copyright They are currently working through the US courts. This is not necessarily a slam dunk for the plaintiffs who claim that AI companies have violated their IP rights.
All for Headnotes
Ross was accused of using Headnotes – summaries of legal decisions – by Westlaw, the Reuters Legal Research Service, to train AI. Ross released AI as a tool for analyzing the documents and executing quests based on questions on all court deposits.
Ross claimed that the use of copyright was legally defender because it was transformative, which means that it was reinstalling the headnotes to serve a noticeably different function or market. In his brief decision, Stephanos Bibas, the judge chairing the case, did not find that the argument was particularly convincing.
Ross, Bibas said in his view, re -installed Westlaw Headnotes in a way that immediately copying Westlaw’s legal research service. The starting platform did not offer new importance, purpose or comments, Bibas identified – undermining the allegation of Ross’s transformative use.
In his decision, Bibas also mentioned Ross’s commercial motives as a reason why the start of the start lost the mark. Ross tried to take advantage of a product that competed directly with Westlaw and without a significant “re -concentration” of IP -protected Westlaw material.
Shubha Ghosh, a professor at the University of Syracuse, who studies the IP law, called it a “strong victory” for Thomson Reuters.
‘The trial will proceed, [but] Thomson Reuters made a brief decision, a victory at this stage of the dispute, “Ghosh said.” The judge also confirmed that Ross had no right to a summary of his defenses, such as fair use and merger. Consequently, the case continues to try with a strong victory for Thomson Reuters. ”
Narrow
Already, at least one set of plaintiffs in another Copyright AI case have asked a court to examine Bibas’s ruling. But it is not yet clear if the previous one will affect other judges.
Bibas’s opinion made a point of discrimination between “Genetic AI” and the AI used by Ross, which did not create content, but simply spits the judicial views that had already been written.
Genetic AI, which is at the heart of copyright against companies such as Openai and Midjourney, are often trained in huge amounts of content from public sources around the web. When fueled by many examples, the genetic AI can create speech, text, images, videos, music and more.
Most companies that develop AI genetically argue that Doctrines of fair use They protect their practice for scraping data and their use for training without offsetting – or even believing – the owners of the data. They argue that they have the right to use any available in the public for education and that their models are valid for transformative projects.
But he does not agree with any copyright holder. Some show the phenomenon known as reflux, where genetic AI creates content that is very similar to the work in which it was trained.
Randy McCarthy, US Patent Lawyer in the law firm Hall Estill, said Bibas’ focus on “market impact on the original project” could be the key to the cases of AI’s rights holders. But he also warned that Bibas’s opinion is relatively close and that it can be overthrown.
“One thing is clear, at least in this case: merely using copyright protected as training data [for] An AI cannot be said to be a fair use per se, “McCarthy told TechCrunch.”[But it’s] A battle in a bigger war and we need to look at more developments before we can explode from this law related to the use of copyright protected materials as AI training data. “
Another attorney TechCrunch spoke to me, Mark Lezama, a partner in Knobbe Martens who focuses on patents, believes that Bibas’ opinion could have wider consequences. It is of the view that the judge’s reasoning could be extended to genetic AI in its various forms.
“The court rejected a fair defense as a legal issue in part because Ross used [Thomson Reuters] Headnotes for the development of a competitive legal system of legal research, “he said.” Although the court hints that this may be different from a situation that includes a genetic AI, it is easy to see a news site that argues that copying its articles for the The training of a genetic AI is no different because the genetic AI uses copyright -protected articles to compete with the News site for users’ attention. ”
In other words, the issuers and copyright owners who take it with AI companies have a light reason to be optimistic after the decision – emphasis small.
