•  
  •  
 

UC Law Science and Technology Journal

Shelby Ponton

Abstract

Should AI companies be allowed to “train” their models on the copy- righted works of others without consent or compensation? Legally, can they? These questions are being litigated in courts across the United States right now. When a resource, such as AI, is engulfed in effective rights of exclusion from a vast array of battling rightsholders, that resource is susceptible to un- derutilization. This phenomenon is referred to as a tragedy of the anticom- mons. This Article highlights how AI is subject to an anticommons weak- ness. If the millions of intellectual property holders, whose intellectual property these AI models are trained on, all see their rights of exclusion be- come effective, it could signal the end of AI before we know it.

As the first Article to shine a light on the anticommons property at the intersection of AI and intellectual property, this narrow focus reveals multi- ple possible solutions to thwart a tragedy of the AI anticommons. Relying on the cross section of traditional entitlements theories and efficiency ration- ales, possible solutions such as private market-based actors and legislatively compulsory licensing emerge. In addition to exposing these pre-existing mechanisms, this Article goes one step further by demonstrating how trans- ferring untapped bankruptcy principles into the AI and intellectual property anticommons is the novel solution this problem needs. Utilizing trusts and channeling injunctions from bankruptcy law to prevent a tragedy of the AI anticommons is a unique approach that allows for a clarification of intellec- tual property entitlements and a reduction in bargain-based transaction costs.

Download

Share

COinS