Online Library

VOX-Pol Blog Post.

Considerable attention has been paid by researchers to social media platforms, especially the ‘big companies’, and increasingly also messaging applications, and how effectively they moderate extremist and terrorist content on their services. Much less attention has yet been paid to if and how infrastructure and service providers, further down ‘the tech stack’, deal with extremism and terrorism. Content Delivery Networks (CDN) such as Cloudflare play an underestimated role in moderating the presence of extremist and terrorist content online as it is impossible for these websites to operate without DDoS protection. This is evidenced by the takedown of a wide range websites such as The Daily Stormer, 8chan, a variety of Taliban websites and more recently the organised harassment site Kiwifarms following refusal of service by Cloudflare. However, it is unclear whether there is any formal process of content review conducted by the company when it decides to refuse services. This article aims to first provide an analysis of what extremist and terrorist websites make use of Cloudflare’s services as well as other CDNs, and how many of them have been subject to takedown following refusal of service. Following this the article analyses CDNs’ terms of service and how current and upcoming internet regulation applies to these CDNs.

VOX-Pol Blog post.

This report treats developments in the violent extremist and terrorist online scene(s) in the 12-month period from 1 December 2020 to 30 November 2021. It accomplishes this by surveying, describing, and integrating the findings of relevant articles and reports produced by academics, thinktanks, civil society, and governmental organisations; high quality media coverage; and the first hand experience and primary research of the authors.
This report treats developments in the violent extremist and terrorist online scene(s) in the 12-month period from 1 December 2020 to 30 November 2021. It accomplishes this by surveying, describing, and integrating the findings of relevant articles and reports produced by academics, thinktanks, civil society, and governmental organisations; high quality media coverage; and the firsthand experience and primary research of the authors.

Online terrorism has been subject to significant change over the past twenty years. An appreciation of the roles of the Internet in terrorism was, like the Web itself, still nascent at the time of 9/11. An early focus on cyberterrorism was nevertheless, even then, giving way to a realisation that the Internet’s affordances were already altering the extremism and terrorism landscape. Some of the earliest work in this area was on extremist and terrorist websites and their purposes.

Extremist and terrorist websites never really went away, they were just overlooked for a decade by researchers and others due to a not unwarranted narrowing of focus to social media platforms and, latterly, messaging applications and adjacent online spaces.

While there is less reliance on websites by extremists and terrorists than there once was, websites remain an important component of the contemporary online extremist and terrorist ecosystem(s) and could re-emerge more strongly with accelerated disruption of extremist and terrorist content and accounts by social media platforms and adjacent services unless providers further down ‘the tech stack’ take more concerted action.

Policymakers have recently expressed concerns over the role of recommendation algorithms and their role in forming “filter bubbles”. This is a particularly prescient concern in the context of extremist content online; these algorithms may promote extremist content at the expense of more moderate voices. In this article, we make two contributions to this debate. Firstly, we provide a novel empirical analysis of three platforms’ recommendation systems when interacting with far-right content. We find that one platform—YouTube—does amplify extreme and fringe content, while two—Reddit and Gab—do not. Secondly, we contextualise these findings into the regulatory debate. There are currently few policy instruments for dealing with algorithmic amplification, and those that do exist largely focus on transparency. We argue that policymakers have yet to fully understand the problems inherent in “de-amplifying” legal, borderline content and argue that a co-regulatory approach may offer a route towards tackling many of these challenges.