COULD THE GOVERNMENT AND SOCIAL MEDIA PROVIDERS COMBAT ONLINE TERRORIST RECRUITMENT?
Last month several U.S. national security officials met with top Internet industry CEOs in Silicon Valley to enlist their help in the surveillance of online terrorist recruitment. Soon after, Google announced two steps in the anti-recruitment direction. One initiative would let non-profit organizations place counter-radicalization ads in response to certain terrorist-related search queries. The other would make counter-radicalization videos easier to find in search results.
Google’s efforts appear to require minimal business burdens and pose no privacy harm. But will they meet the government’s expectations?
According to news reports on the Silicon Valley summit, the government asked the Internet giants for much more than adjustments to their advertising and search features. Among other things, the government reportedly wanted help monitoring “social media recruitment patterns.” The apparent goal was to disrupt the kind of Islamic State radicalization that led to the mass-shooting at a San Bernardino public health facility in December.
The technology to monitor social media communications already exists. If law enforcement were to give the social media providers (e.g. Google, Facebook, Twitter) a set of key words and communication patterns associated with terrorist recruitment, each provider could monitor its traffic for these identifiers using a scanning algorithm and a system of alerts. A communication that triggers an alert would be referred to law enforcement for investigation. Law enforcement could then present the evidence to a court. If the judge makes a finding of probable cause he or she could sign an order authorizing the agents to further the scope of the investigation, and potentially, intercept the suspect’s communications. The surveillance may lead to an arrest on grounds that the suspect violated the federal prohibition against providing material support to terrorism.
In Subsentio’s view, implementing this pattern-based investigative analysis requires attention to certain legal issues. To begin with, social media providers have no obligation to build technology in their networks to assist law enforcement. The most relevant statute is the Communications Assistance for Law Enforcement Act, or CALEA. The CALEA statute requires telephone companies, cellular radio carriers, two-way interconnected VoIP providers and broadband access providers to install target-centric communications surveillance solutions in their networks. But the statute exempts providers of “information services” such as Google Plus and Facebook. Congress would have to amend the law before it could govern social media hosts.
The next issue is whether social media providers may lawfully apply algorithms to the content of their communications traffic to detect patterns of end user behavior. The answer is yes. In fact, they already use such technology for purposes of Internet advertising. They also use another software solution to identify and report child pornography to the National Center for Missing and Exploited Children.
Beyond that, the providers meddle in user content even more intrusively. They implement “take-downs” of postings that infringe on copyrights and remove videos or other speech deemed offensive under their “terms of use” policies. Some, including Facebook, ban the use of their sites for the private sale of guns and illegal drugs.
The most novel legal issue posed by the pattern-analysis approach is how the communication patterns would be incorporated in a showing of probable cause. Specifically, what pattern of emails, instant messages, or VoIP communications would constitute probable cause that a suspect is a terrorist recruiter? Would the judge look for certain pro-terrorist messaging? A certain number of emails, instant messages or calls containing such messaging? Pro-terrorist messaging to a minimum number of potential recruits?
When material support for terrorism consists of fund raising, arms dealing, or passport forging, amassing evidence of probable cause in the terrorist investigation is no different from performing the same task in a regular criminal investigation. But when the material support for terrorism consists of online recruiting for a terrorist group, the targeted evidence is speech. And when the evidence that purports to justify the wiretap order consists solely of algorithmic pattern analysis, as opposed to traditional detective work, a software bug may trigger a wiretap of innocent speech.
If the government and the communications industry were to cooperate to monitor online radicalization, their engineers could do the job. But their lawyers would need to ensure it’s done right.