America’s Next “Stop Model!”: Model Deletion
Famed American model and television personality Tyra Banks once wrote, “Love me or hate me I promise that it will never make or break me…<3.”1 Fortunately, the same cannot be said for harmful artificial intelligence (AI) and machine learning (ML) models.2 Indeed, harmful AI and ML models can and should be broken and more specifically, deleted.3
It could be marketing materials or privacy policies that mislead users about whether uploaded photos are used to train facial recognition models,4 or a mobile application that fails to get verifiable parental consent to collect minors’ data and use it to train nutrition models.5 Businesses causing algorithmic harm should face deletion—effectively destruction6—of significant portions of their AI and ML related work products, including datasets, algorithms, models, and materials affected by the illegally acquired data.
Continue Reading
Jevan Hutson & Ben Winters
Jevan Hutson, Associate Attorney at Hintze Law PLLC, Seattle, WA. University of Washington School of Law, J.D. 2020. The views expressed in this article are his own and do not necessarily reflect the positions of any current or former employer or client; Ben Winters, Senior Counsel at Electronic Privacy Information Center (EPIC) and lead of EPIC’s AI and Human Rights Project, Washington, D.C. Benjamin N. Cardozo School of Law, J.D. 2019.