Thompson Reuters wins in an early battle against the court over artificial intelligence, copyright and fair use

On Tuesday, the American Dilayer Provincial Court, Stefanus Pipas, issued a partial brief sentence in favor of Reuters in Thompson in a lawsuit against copyright against it against it against it. Ross intelligenceBeginning of the legal operation of artificial intelligence. It was presented in 2020, and it is one of the first cases that will deal with the legitimacy of artificial intelligence tools and how to train it, and often used data for copyright from another place without a license or permission.

Similar lawsuits against Openai, microsoftAnd other artificial intelligence giants currently wandering through the courts, and they can reach similar questions about whether artificial intelligence tools can demand the defense of “fair use” for the use of copyrights protected.

In a statement he made freedom Written by Thomson Reuters, spokesman for Jeff McCoy, the company said:

We are pleased that the court granted a brief ruling in our favor and concluded that the content of the Westlaw editing that was created and maintained by our lawyers, is protected by copyright and cannot be used without our approval. We were not copying our content “fair use”.

However, as the judge indicated, this case included “non -degenerative” artificial intelligence, not the AI ​​Gynecological Tool like LLM. Ross closed in 2021, Call the “false” lawsuit But saying that he was unable to collect enough financing to continue working while falling into a legal battle.

As mentioned earlier by WirelessToday, Judge Pepas He wrote in his decision“None of the potential Russian defenses carry water” against the accusations of violating copyright, and eventually rejected Russian defense for fair use, and to rely heavily on a Russian use factor for copyrights that affected the market for the value of the original work by building Direct competitor.

Thompson Reuters filed a lawsuit against Ross’s use of the search engine in Westlau. Westlaw is a large number of copyrights (such as legal decisions) but also permeates its own content. For example, WestLaw Headnotes – summaries of law points written by human editors – is a signature feature aimed at making Westlaw subscription very expensive to lawyers.

When building a legal search engine, on Ross, illustrations and heads “to digital data on relations between legal words for feeding in artificial intelligence”, Bibas wrote. The referee describes how, after Thompson Reuters rejected her attempt to license the Westlaw content, Ross turned into another company, Legalese, and bought 25,000 loose notes of questions and answers written by lawyers using the Westlaw Headnotes that I used for training data.

Ross Andrew Arouda CEO Claim Westlaw data was “noise” and that its tool “aims to identify answers and extract them directly from the law using machine learning.” However, after comparing “the extent of the similarity of each of the 2,830 loose memorandums, the main elements, and judicial opinions are, one by one,” the judge said that the evidence for the actual copies “is very clear so that no reasonable jury can To find otherwise. “