Click here to read GNET's latest report GNET Survey on the Role of Technology in Violent Extremism and the State of Research Community-Tech Industry Engagement

Defining the Delta Between Hashed Terrorist Content and All Terrorist Content

Defining the Delta Between Hashed Terrorist Content and All Terrorist Content
17th May 2021 Brett Raffish
In Insights

Terrorists and violent extremists (“VE”) have “exploit[ed] digital platforms” to share intelligence, recruit, and disseminate ideology. The Global Internet Forum to Counter Terrorism (“GIFCT”) sought to remedy this problem by providing its members—Facebook, Twitter, Microsoft, and other platform giants—access to a shared database of hashed or “tag[ged]” terrorist videos and images allowing members to quickly and uniformly remove such terrorist and VE content. Despite GIFCT’s efforts, users continue to exploit platforms to advance their violent ends.

The theoretical delta between terrorist and VE content hashed by GIFCT and all terrorist and VE content posted on members’ platforms may be attributed, at least in part, to GIFCT’s reliance on the U.N. Security Council consolidated sanctions list (“the U.N. list”). Researchers Chris Meserole and Daniel Byman suggested that any designation list adopted by platforms be (1) “unbiased[,]” (2) “global[,]” and (3) “updated in real-time[.]” When measured against Meserole and Byman’s framework, the U.N. sanctions list likely falls short under all three criteria.

In this post, I greatly expand on Meserole and Byman’s analysis of the U.N. list detailed in their 2019 Royal United Services Institute white paper. I agree with Meserole and Byman’s conclusions and explain in this post how the U.N. list: (1) is geopolitically skewed, limiting GIFCT’s global reach; (2) inappropriately operationalises terms like “terrorist” and “violent extremist,” creating systemic bias by permitting GIFCT to capture a subset of terrorist and VE organisations; and (3) may not be timely amended in practice in light of the absence of organisations and individuals involved in recent attacks. These shortcomings may significantly limit GIFCT’s ability to effectively “[p]revent terrorists and violent extremists from exploiting digital platforms[.]”

Geopolitical Skew

The U.N. list is presently geopolitically skewed, limiting GIFCT’s ability to reach terrorists and violent extremists operating worldwide but who are absent from the U.N. list. Indeed, a designation list which “only covers one country or region . . . will be of limited use to social networks and file-sharing platforms with a global userbase.” Twitter, YouTube, and Facebook unquestionably possess global membership. Yet, the U.N. list is undeniably regional as it focuses primarily on activity in Southwest Asia and Africa. Except for the Democratic People’s Republic of Korea, U.N. sanctions are presently concerned with Somalia, Iraq, Sudan, Libya, Yemen, Mali, South Sudan, Guinea Bissau, the Central African Republic, and the Democratic Republic of Congo. Other committees target Islamist terror groups, such as Islamic State (IS), Al-Qaeda, and the Taliban, which operate globally but are tethered to the global regions highlighted above.

Groups that some have labeled terrorists or violent extremists located outside of the aforementioned regions are noticeably absent from the list. These groups include, but are not limited to, Revolutionary Struggle, the Continuity Irish Republican Army, Shining Path, Aum Shinrikyo, Hezbollah, Azov, The Base, New People’s Army, the Russian Imperialist Movement, Atomwaffen Division, Combat 18, Euskadi Ta Askasuna, and Blood & Honour. Because GIFCT’s database flows from the U.N. list, it may miss terrorist and VE content promulgated by groups located in regions beyond those captured by the list.

A Definitional Discrepancy

As Meserole and Byman appear to allude to in their paper, a discrepancy exists between the mechanisms the U.N. employs to determine its list’s composition and GIFCT’s intended targets. GIFCT targets terrorists and violent extremists. However, GIFCT’s mission begs the question: who are terrorists and violent extremists? The U.N. fails to definitively answer this question as it does not appear to own or apply a “universal” or widely accepted definition of either terrorism or violent extremism. Recent Security Council resolutions addressing groups like Al-Qaeda and IS appear to explicitly and implicitly apply elements of a working definition of terrorism laid out in Resolution 1566, passed in 2004. Although unclear and assuming without concluding that the U.N. is relying on the 1566 definition to award sanctions for terrorism, meeting the definition’s criteria does not automatically trigger sanctions, even if sanctions are permissible. In other words, being a terrorist does not mean one will be formally sanctioned, plausibly creating a gap between all terrorists who fit the U.N.’s criteria but are never sanctioned and listed with the sample of international terrorists who are sanctioned and whose content can be hashed by GIFCT.

Whether this discrepancy makes the consolidated list itself systemically biased is unclear at best. Meserole and Byman define “unbiased” as “all entities that meet a particular definition are equally likely to appear on the list regardless of ideology or identity.” Assuming that some terrorist groups meet the 1566 definition but are never sanctioned, one could plausibly conclude that the U.N. list is biased. On the other hand, however, the sanctions list is mechanically and fundamentally different than a traditional terrorism designation list because the sanctions list demands that a group or individual do more than engage in terrorist acts, which may be sufficient to earn a spot on a terrorism designation list. Indeed, acts must be severe enough for international sanctions and condemnation and that sanctioning must happen before an individual or organisation earns a spot on the consolidated list. Therefore, even if the U.N. list is not itself systemically biased, it may, for reasons described above, create systemic bias in GIFCT’s database by only permitting GIFCT to taxonomically hash content related to sanctioned or sanctionable terrorists, who may, in theory, represent a subset of all terrorists. Furthermore, the ambiguous U.N. definition may simply be asynchronous with other definitions of terrorism that may be more ‘true’ or ‘correct’ than the 1566 definition, evincing that GIFCT is conceivably missing terrorist and VE groups that do not meet U.N. sanctions criteria but may nevertheless meet some definition of VE and terrorism. Finally, the 1566 definition focuses on groups or individuals who have acted, removing the possibility that GIFCT preemptively hash content belonging to terrorist groups who promulgate or promote violence, but who have not committed a “criminal act[][.]”

Relatedly, the consolidated list largely neglects the other half of GIFCT’s mission—thwartingviolent extremists from exploiting digital platforms”—creating systemic bias in GIFCT’s database by focusing on terrorism and leaving out violent extremism. The Security Council has considered violent extremism “conducive to terrorism[,]” but has neglected to say whether violent extremism can be sanctioned in the absence of terrorism. Assuming, in the U.N. Secretary-General’s terms, that “[v]iolent extremism encompasses a wider category of manifestations[,]” GIFCT’s database may conceivably fail to capture scores of content belonging to violent extremists who are not terrorists.

Timely Amendments

Finally, the U.N. list can be “updated in real-time[,]” but does not appear to track some global events. The Security Council “may meet whenever peace is threatened.” However, the absence of some violent groups from the list may evince that the U.N. list may fail to keep pace with the precise groups and organisations that GIFCT may want to capture. For example, members of the Russian Imperial Movement have fought in Syria and Libya in 2019 and 2020, respectively. Furthermore, at least one known member carried out multiple violent attacks in Sweden between 2016 and 2017, earning the organisation a spot on the US State Department’s terrorist designation list. Despite the organisation’s ties to terrorist activity, the organisation is absent from the U.N. list. Similarly, members of Atomwaffen Division and its global offshoots were “linked to” several violent acts in the United States between 2017 and 2018, and “plots in the United States and Europe, including death threats made against Belgian MEP Guy Verhofstadt and planned attacks on Las Vegas establishments, including a synagogue and gay bar.” Again, despite the violent organisation’s global presence, it has skirted formal U.N. sanctioning and remains off the U.N. list. Thus, although the U.N. list can be timely updated, it has not been in the wake of some violent acts promulgated by members of alleged terrorist or violent extremist organisations.

GIFCT has instituted CIP to address “emerging and active terrorist events and assess any potential online content produced and disseminated by those involved in the planning or conducting of the attack.” However, GIFCT has deployed CIP judiciously, activating it in response to only a handful of violent incidents. In other words, CIP is not designed to capture smaller-scale violent incidents occurring globally, multiple times a day. Thus, for the foregoing reasons, the U.N. list stymies GIFCT’s ability to address a complex, evolving threat landscape.

Moving Forward

The foregoing illustrates that the U.N. list may hinder GIFCT’s ability to effectively capture terrorist and VE content and fulfill its preventative mission. How GIFCT addresses or remedies these shortcomings is a more complex inquiry outside of this post’s ambit. However, the author broadly encourages GIFCT to consider the feasibility of crafting its own designation list by drawing on widely accepted definitions of terrorism and violent extremism, and to consider overlaying existing designation lists to address some of the U.N. list’s shortcomings detailed above.