Online Library

The French Digital Council has been given a mandate to examine Article 9 of the bill on scaling up
counter-terrorism provisions. These provisions amend Article 6 of the Act of 21 June 2004 on confidence
in the digital economy (LCEN) by providing for the administration to block websites containing speech
and/or images constituting incitement to terrorism or in defence of terrorism. They also extend the scope
of the notification tools that technical service providers are bound to use.
The French Digital Council has held fifteen interviews with terrorism experts (sociologists, journalists and
association representatives), specialised lawyers and judges, civil society representatives, intelligence
agency members and digital professionals (full list available in the appendix) to be able to deliver as fully
informed an opinion as possible.

This paper contributes to platform-comparative research through a case study of Twitter/X, Telegram, and Gettr in the context of German far-right social movements. It introduces the conceptual framework of platform opportunity structures to examine how platforms enable or constrain far-right mobilization. Using community analysis of sharing networks among the same pool of far-right social movement actors, the study explores how technological affordances, governance models, ownership and branding, and user bases and cultures shape platform-specific networking patterns. The findings reveal Telegram as a central platform for the most radical and active communities; Twitter as a site where anti-elite journalists and politicians are salient; and Gettr as a platform to connect to the U.S. far right. While anti-elite and COVID-related conspiracist figures exert influence on all platforms-particularly on Telegram, the prevalence of AfD politicians, pseudonymous amplifiers, and transnational ties to the U.S. far right – especially after Elon Musk’s acquisition of Twitter – are emergent factors. As Twitter shifts toward ‘alternative’ platforms like Telegram and Gettr, offering minimal moderation and security and even ideological branding, this article adds to understanding how dynamic platform opportunity structures shape far-right mobilization online.

The purpose of this article is to inform the debate about strategies and options for countering online radicalization within the U.S. domestic context. Its aim is to provide a better understanding of how the Internet facilitates radicalization; an appreciation of the dilemmas and tradeoffs that are involved in countering online radicalization within the United States; and ideas and best practices for making the emerging approach and strategy richer and more effective. It argues that online radicalization can be dealt with in three ways. Approaches aimed at restricting freedom of speech and removing content from the Internet are not only the least desirable, they are also the least effective. Instead, government should play a more energetic role in reducing the demand for radicalization and violent extremist messages—for example, by encouraging civic challenges to extremist narratives and by promoting awareness and education of young people. In the short term, the most promising way for dealing with the presence of violent extremists and their propaganda on the Internet is to exploit their online communications to gain intelligence and gather evidence in the most comprehensive and systematic fashion possible.

This article is concerned with the distinction between acceptable race talk in social media and organised, extreme or ‘frozen’ racism which is considered hate speech and removed. While in the literature this distinction is used to point to different variants, styles and mutations of racism, in social media platforms and in European regulatory frameworks it becomes policy. The empirical part of the article considers this distinction drawing upon a series of posts following a stabbing incident in a small Irish town, which organised Twitter accounts sought to connect to terrorism. The empirical analysis examines the tweets of those accounts and the comments left on the Facebook page and website of one of the main Irish online news outlets. The analysis shows few if any differences between the two, concluding that there is a blending of supremacist and everyday, ambient racist discourses. This blending indicates the operation of a transnational contagion, given the shared vocabularies and discourses. It further problematises the distinction between ‘illegal hate speech’ and ‘acceptable race talk’, and throws into question the principle underlying both the policies of social media as well as the European efforts to de-toxify the digital public sphere.

Against the backdrop of the COVID-19 pandemic, anti-lockdown movements sparked concerns about online extremism escalating into real-world violence. This study takes the Australian anti-lockdown movement to explore extremist discourse in Australia, as it contained several, overlapping and interacting extremisms. This study explores the labelling of these protests as extreme and examines the interplay between individuals, ideologies and discourse. Understanding the definitions and labels of extremism is fundamental for identifying and countering potential threats, as it distinguishes harmful rhetoric from extremist discourse and enables precise, informed interventions. A qualitative close reading revealed that while Australian anti-lockdown discourse contained crisis narratives and group antagonisms, it did not consistently meet the structural definition of extremism. This is partly due to the problem of coding violence, which lies in the lack of a clear comparative threshold to determine whether a discursive claim, rhetoric, or cultural reference signifies violence or violent intent. While not meeting the structural definition of extremism, there was evidence of specific structural attributes that logically precede extremism and the emergence of violence. By reconceptualizing extremism as a temporal dynamic, this research explores the relationship between discursive and violent extremism, shedding light on the complexities of extremist movements and their social impact.

Digital platforms were slow to build robust teams to counter threat actors, but today, many of those corporate teams have robust processes, specialized tools, and innovative approaches to countering highly adaptive adversaries. They operate in a tremendously dynamic environment where their adversaries can innovate at low cost, primarily because of the nature of the digital “terrain” where the conflict occurs. And while the actions these teams take are not kinetic, as those sometimes utilized in geopolitical conflict, the cat-and-mouse game between Trust & Safety teams and threat actors online suggests lessons that are increasingly relevant to the national security community. This article explores five factors that were key to facilitating innovation in Facebook’s approach to countering the Islamic State—and that I argue are more generalizable. They are: people, organization, legitimacy, tools, and collaboration. It also identifies lessons that can be learned from that experience. For example, we did not prioritize using a particular technology or focus experimentation in some bespoke “innovation center.” Rather, we succeeded because we were made responsible for a critical mission, were unencumbered by past process, and had the right team structured to reduce external dependencies for innovation. Basic technological innovation can occur in an ivory tower, but applied innovation requires proximity to real-world missions. You cannot expect dramatic innovation without failure and iteration in an environment of real responsibility. Fundamentally, that means that innovation requires accepting risk. The structures and incentives of Silicon Valley cannot and should not simply be grafted on to our national security infrastructure. The rewards and costs of failure are different. But military organizations should shoulder the risks associated with innovation and study the lessons of corollary efforts in Silicon Valley and the private sector more broadly.