•  
  •  
 

Cardozo Law Review

Abstract

The rapid proliferation of generative artificial intelligence (“AI”) applications in recent years has sparked a global “AI frenzy.” Yet, the rise of new AI technologies has also raised concerns about copyright infringement, particularly regarding whether generative AI developers should be held liable for using copyrighted works as training inputs and generating outputs that compete with original works. In response, AI developers have invoked a long line of technological fair use precedents to defend against any potential copyright infringement claims. The purpose of this Article is to offer insights into whether generative AI is a transformative technology that deserves fair use protection, or simply a blatant thief of authors’ fruits of labor. This Article first analyzes activities that may constitute infringement at both the input and output stages of the AI content generation process from a technical perspective. It then reviews the ongoing generative AI copyright infringement cases to identify which infringement claims litigants and courts consider most likely to succeed. Since plaintiffs’ claims of input and output infringement so far seem likely to survive the pleading stage, this Article subsequently explores whether and how the fair use doctrine—particularly in light of precedents concerning non-expressive uses—may justify AI defendants’ infringing activities. Evaluating the fairness of input and output copying separately under the four statutory fair use factors, this Article concludes that while AI defendants’ fair use defense is viable for input infringement, it is far less likely to prevail in the context of output infringement.

Disciplines

Intellectual Property Law | Internet Law | Science and Technology Law

Download

Share

COinS