Is it broken?

https://scholarship.claremont.edu/cmc_theses/3463/

Copyrights and Wrongs: Evaluating Copyright Law's Adaptability to Generative AI

The legality of generative artificial intelligence (GAI) within the realm of copyright law remains uncertain. GAI models, by ingesting and training on large amounts of data, develop the ability to generate novel expressive outputs such as text and images. The quality of these outputs is directly related to the quality of their inputs. Consequently, developers often use copyrighted works as a primary source for training, recently resulting in lawsuits from copyright holders alleging infringement. Developers defend themselves by claiming that their actions constitute fair use, but the murky nature of fair use casts doubt on whether GAI models can definitively be classified as such. Moreover, reliance on fair use doctrine may lead to undesirable outcomes, potentially impacting GAI innovation, human creativity, or both, thereby undermining the constitutional goal of copyright to `promote the Progress of Science and useful Arts.' This thesis evaluates whether current copyright law can adapt to GAI challenges and maintain its aim of balancing protections and incentives. It also explores the necessity of amendments to copyright law to effectively address GAI, envisioning possible alternatives.