Author: Joe Karaganis (The American Assembly, Columbia University)
The practice of notice and takedown under the DMCA has changed dramatically in the last five or six years, driven by the adoption of automated notice-sending systems. As these systems became common, the number of takedown requests to many services skyrocketed, quickly overwhelming human vetting at the targeted services. Because of the liability risk associated with ignoring a DMCA request, most targeted services responded with “DMCA+” measures for managing the takedown process on the new, much larger scale, ranging from blanket takedown, to algorithmic triage, to content filtering.
Increasingly, the online regulation of speech passes through such systems, subject to little human intervention or verification and a relatively poor record of accuracy. Our work traces this history and evaluates the reliability of automated procedures, based on interviews with service providers and coding of Google Search notices.