The legal field surrounding artificial intelligence (AI) is an emerging creature, but generative AI, a stand-in for the latest “chatbots” that have exploded in popularity recently, has been a source of legal development since 2023.
At its core, generative AI leverages computer systems to gather extensive data from across the internet, often incorporating external intellectual and economic contributions. As a result, original authors have filed scores of infringement claims against AI companies like Google, Meta, Microsoft, and OpenAI, arguing they have wrongfully “ingested” their original works in developing large language models.[1] AI companies have defended their actions by asserting the “Fair Use” exception to the Copyright Act of 1976.[2] Many of these cases remain active and the fair use issue unresolved.
Fair use is an affirmative defense to copyright infringement, and the defendant infringer bears the burden of proving the defense. Courts consider the following factors in deciding whether a use of intellectual property is considered fair use: (1) the purpose and character of the use, (2) the nature of the copyrighted work, (3) the amount and substantiality of the portion used, and (4) the effect of the use upon the potential market for the copyrighted work.[3]
In these cases, the third element appears to be the most problematic. Language models are trained similar to how a human brain is trained to master a new skill—by constant repetition and exposure to various examples of how to do something correctly. Identifying the precise quantity of a single original work used for training a language model when the process involves scraping thousands of other works across a multitude of sources in an instant remains difficult to prove. In addition, AI companies are often hesitant to reveal their algorithms in fear of losing competitive advantage, thus creating a black box problem.
As these cases remain pending, legislators have recognized the need for further regulation. In October 2023, President Biden signed Executive Order 14110 into law which gives the Executive Branch authority to define new goals for AI regulation. In April 2024, Congress proposed the Generative AI Copyright Disclosure Act, which would require companies to submit a notice of the identity and URL address of any copyrighted work used in training data to the Copyright Office at least 30 days before releasing new generative AI models. This law has yet to go into effect.
Until comprehensive legislation is adopted and many of these active cases resolved, the legality of the use of generative AI against the obvious need for IP protection remains to be seen.
If you have any questions about these active cases or how generative AI applies to your business, please contact your Masuda Funai relationship attorney or any member of Masuda Funai's Litigation Practice Group.
[1] Andersen v. Stability AI Ltd., No. 3:23-cv-00201 (N.D. Cal.); Tremblay et al. v. OpenAI, Inc. et al., No. 3:23-cv-03223 (N.D. Cal.); Authors Guild v. OpenAI, Inc., No. 1:23-cv-08292 (S.D.N.Y.); Getty Images (US), Inc. v. Stability AI, Inc., No. 1:23-cv-00135 (D. Del.); The New York Times Co. v. Microsoft Corp., No. 1:23-cv-11195 (S.D.N.Y.)
[3] Google LLC v. Oracle America, Inc. (2021) 593 U.S. 1, 2.
Masuda Funai is a full-service law firm with offices in Chicago, Detroit, Los Angeles, and Schaumburg.
©2024 Masuda, Funai, Eifert & Mitchell, Ltd. All rights reserved. This publication should not be construed as legal advice or legal opinion on any specific facts or circumstances. The contents are intended solely for informational purposes and you should not act or rely upon information contained herein without consulting a lawyer for advice. This publication may constitute Advertising Material.