In one of the first lawsuits to allege that generative AI companies violate the U.S. Copyright Act by using copyrighted works to train machine learning models, Judge Stephanos Bibas of the Delaware Circuit Court recently denied the majority of issues raised in cross motions for summary judgment filed by plaintiff Thomson Reuters and defendant Ross Intelligence Inc. The court declined to issue a dispositive ruling on the hot-button question of whether the fair use doctrine protects generative AI companies that use copyrighted materials to train their programs.
Thomson Reuters (owner of Westlaw) sued Ross Intelligence, a legal-research generative AI startup, in May 2020, alleging that Ross was liable for both copyright infringement and tortious interference with contract. The allegations against Ross stem from its endeavor to create a search engine that uses machine learning and artificial intelligence to provide answers to commonly asked legal questions.
In need of material to train its generative AI, Ross attempted to obtain a license to use Westlaw. When Westlaw turned Ross away, it asked third-party legal research companies to provide it with legal material — much of which those legal research companies obtained from Westlaw. Thomson Reuters contends that Ross copied large portions of Westlaw’s Headnotes and Key Number System.