Copyright Law & Artificial Intelligence: Is Training AI With Other’s Data Fair Use – Professor Mark Lemley (Stanford Law)

Do AI developers need permission to use copyrighted works and other IP before using it to train artificial intelligence?

Professor Mark Lemley of Stanford does not think so. He believes using copyrighted works to train AI should fall under the fair use exception to copyright law.

Professor Lemley is the Director of the Stanford Program in Law, Science and Technology, an author of seven books and more than 130 articles on intellectual property, antitrust and related areas of the law. He is also a co-founder of Lex Machina and most recently Of Counsel to Lex Lumina, a boutique IP law firm.

Professor Lemley argues that AI companies should be permitted to use copyrighted works to train AI models without first getting permission from owners because of the benefits AI will yield and the impossibility of tracking down millions of copyright owners to get permission.

He also believes that it is a fair use for AI developers to use works protected by intellectual property laws to train artificial intelligence models because such a use is transformative and the more data available to the AI, the more accurate it will be.