Judge rules against tech firm in first major AI copyright case

A US federal court has ruled against an AI startup in a high-profile copyright case.

The Delaware court, led by Judge Stephanos Bibas, ruled that the now-defunct ROSS Intelligence violated copyright law by using headnote summaries, created by Thomson Reuters’ editors, to train its AI-powered legal research product without permission.

The case, which began in 2020, saw the judge reject ROSS's "fair use" argument, ruling its use of the materials was commercial and harmed Thomson Reuters' market by creating a competing product.

After the failure of the fair use argument, the judge also upheld that the headnotes are original works, crafted through a creative editorial process, and thus fully protected under copyright law.

This ruling is seen as a major precedent in the ongoing clash between AI startups and intellectual property rights.

It comes two months after the UK government opened a consultation, which will run until 25th February, to determine what constitutes AI copyright infringement.

The outcome of this case is expected to influence many other AI copyright lawsuits currently pending in US courts.

Companies like OpenAI, Anthropic, Stability AI and MidJourney are facing a multitude of legal issues over training data usage.

Music publishers sued Anthropic in 2023 for allegedly using lyrics from 500 songs without authorisation. It reached a partial settlement last month, agreeing to add safeguards that will prevent its AI model from reproducing copyrighted lyrics, but the case remains open.

Meanwhile, OpenAI has been on the negotiating table to secure licensing deals with publishers since mid-2023. It recently penned a deal with News Corp, TIME, The Atlantic and others to address copyright concerns.