Document Type


Publication Date



This very second, a battle between robots and pirates is being waged online. Pirates are stealing content from copyright holders and uploading it to various websites. Robots are crawling the Web, searching for pirated content. When a robot encounters pirated content, it is programmed to attack — either by reporting back to the copyright holder, or by going straight to the source and requesting that the material be removed. Sometimes the pirates fight back, re-posting the content online soon after it is taken down or posting newly infringing content. The cycle continues, and the battle rages on.

Generally speaking, “robots” like the one described above operate in a certain sequence. First, they scan the Internet for certain keywords or other indicia of copyrighted content. Once the robot locates content it has identified as copyrighted, it informs the copyright holder, who then decides how to react. The copyright holder may investigate the lead further, and, if the content is indeed found to be infringing, may send a request to the Internet service provider (“ISP”) to remove the infringing content. This action is known as a “takedown” or “noticeand-takedown” request.

The robot, however, can also initiate what is known as a “robo-takedown” by contacting the host site directly and requesting that the content be removed. By definition, robo-takedown requests are issued pursuant to a largely — if not fully — automated review process, meaning that there is no human actor reviewing the content to verify that it is in fact copyrighted before requesting that it be removed. Since each takedown request can identify an infinite number of sources pointing to content the algorithm has identified as infringing, this process can sometimes target content that is, in fact, noninfringing.

This Note argues that automated notice-and-takedown processes are in fact the most efficient means available to deal with the high volume of infringing content in the digital world, but that the counter-notification procedure needs to be strengthened to provide more protection to noninfringing users in the event that their content is removed. Further, the process of scanning for and identifying infringing content must be improved to lessen the incidence of overbroad requests. Part I of this Note provides background on the Digital Millennium Copyright Act (“DMCA”) and the evolution of the notice-and-takedown procedure. Part II outlines the problems with the current state of the notice-and-takedown environment, provides an overview of the data available on takedown requests, and presents an empirical analysis of all known litigation arising from user pushback on such requests in an attempt to shed new light on the amount of noninfringing content that is actually targeted. The overarching aim of Part II is to help answer the question of whether takedown requests, generally speaking — but with an emphasis on robo-takedown requests — tend to be abusive in practice, or if instances of overbroad requests are the exception rather than the norm. Finally, Part III suggests ways in which automated notice-and-takedown processes can be refined to strike a better balance between copyright holders, ISPs, and the general public.


Winner of the 2016 Andrew D. Fried Memorial Prize.