Abstract
The National Security Agency's bulk data collection programs dig closed in 2013 suggest a new model of surveillance. This "predictive surveillance" model will apply the emerging field of predictive analytics to the vast datasets in the hands of third parties. Unlike traditional surveillance, predictive surveillance will not begin by targeting individuals based on particularized suspicion. Instead, predictive surveillance will first collect and analyze all available data to find patterns that could predict future events. In light of the NSA's emphasis on data analysis and its existing stores of telephone and Internet communications metadata, it seems inevitable that the government will eventually advocate for predictive surveillance in antiterrorism investigations. This Article contends that the existing framework of surveillance regulation cannot adapt to predictive surveillance because the existing framework presumes that surveillance begins with targeting. The Article next assesses whether predictive surveillance could comply with the Fourth Amendment. After rejecting arguments that the Fourth Amendment should not apply to information gathered from third parties or in public spaces, the Article proposes a narrow basis for authorizing predictive surveillance. The initial data collection would be reasonable under the Supreme Court's balancing approach in domestic security investigations and under the Court's approval of suspicionless searches in its special needs cases. This approach, however, would only uphold surveillance used for antiterrorism investigations, rather than ordinary law enforcement purposes. Finally, the Article proposes a new regulatory framework that postpones the assessment of particularized suspicion, requires prior judicial approval of the initial data collection and analysis, limits use of the database to antiterrorism investigations, and imposes substantial oversight and transparency.