Thomson Reuters wins a first legal battle on AI, copyright and fair use

Thomson Reuters wins a first legal battle on AI, copyright and fair use

Tuesday, the judge of the American district court of Delaware, Stephanos Bibas Ross IntelligenceA legal AI startup. Placed in 2020, this is one of the first cases that will treat the legality of AI tools and how they are trained, often using data protected by copyright extinguished elsewhere without license or authorization.

Similar prosecution against OPENAI, MicrosoftAnd other AI giants are currently making their way to the courts, and they could be summed up with similar questions as to whether or not the “equitable use” defense can claim the ” Use of material protected by copyright.

In a declaration given to The penis By the spokesperson for Thomson Reuters, Jeff McCoy, the company said:

We are happy that the court has rendered a summary judgment in our favor and conclude that the editorial content of Westlaw created and maintained by our lawyer publishers, is protected by copyright and cannot be used without our consent. The copy of our content was not “fair use”.

However, as the judge noted, this case involved an “non -generative” AI, not a generative AI tool like an LLM. Ross closed in 2021, Call the “False” trial But by saying that he was unable to raise enough funding to continue while taking a legal battle.

As indicated above by CableToday Judge Bibas written in his decision“None of Ross’ possible defenses contains water” against accusations of copyright violation, and finally rejected the defense for fair use of Ross, based strongly on the factor of the way in which the Use by Ross of the material protected by copyright affected the market of the value of the original work by building a direct competitor by Ross.

Thomson Reuters continued on Ross use of its Westlaw search engine. Westlaw indexes a large part of equipment which is not corrected (such as legal decisions) but which also interprets it with its own content. For example, Westlaw craft notes – which are summaries of law points written by human publishers – are a signature characteristic intended to make the subscription very expensive Westlaw attractive to lawyers.

By building a legal research engine, Ross has transformed annotations and host notes “into digital data on relations between legal words to feed its AI”, wrote Bibas. The decision describes how, after Thomson Reuters rejected his attempted content leave of Westlaw, Ross turned to another, legal company and bought 25,000 bulk memos of questions and answers written by lawyers using the Westlaw Aras-à-Tête notes that she used for data training.

CEO of Ross Andrew Arruda claimed Westlaw data has been “added” and that its tool “aims to recognize and extract responses directly from the law using automatic learning”. However, after having “compared to what extent each of the 2,830 memo questions in bulk, head notes and judicial opinions is one by one”, the judge declared that proof of the real copy was “so obvious that No reasonable jury could find the opposite ”.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *