Online Library

In 2020, OpenAI developed GPT-3, a neural language model that is capable of sophisticated natural language generation and completion of tasks like classification, question-answering, and summarization. While OpenAI has not opensourced the model’s code or pre-trained weights at the time of writing, it has built an API to experiment with the model’s capacity. The Center on Terrorism, Extremism, and Counterterrorism (CTEC) evaluated the GPT-3 API for the risk of weaponization by extremists who may attempt to use GPT-3 or hypothetical unregulated models to amplify their ideologies and recruit to their communities.

Recent elections in Europe have demonstrated a steady rise in the success of right-wing populist parties. While advancing an anti-immigration agenda, these parties have been adamant to distance themselves from ‘right-wing extremism’. This article analyses a sample of tweets collected from the Twitter accounts of the German AfD, Identitarian Movement and the Autonomous Nationalists by employing frame analysis. We conclude that the frames of far-right actors classified as extremist, New Right, and populist in fact converge and we discuss our findings in the context of related case studies in other European countries.

There are increasing calls to curtail or revoke Section 230, a 1996 law that protects internet companies from most lawsuits related to user-generated content. Critics of Section 230 say it discourages vigilant self-regulation. Proponents counter that the law has fostered free expression and innovation. Our report concludes that Section 230 ought to be preserved but significantly amended.

Facebook facilitates more extensive dialogue between citizens and politicians. However, communicating via Facebook has also put pressure on political actors to administrate and moderate online debates in order to deal with uncivil comments. Based on a platform analysis of Facebook’s comment moderation functions and interviews with eight political parties’ communication advisors, this study explored how political actors conduct comment moderation. The findings indicate that these actors acknowledge being responsible for moderating debates. Since turning off the comment section is impossible in Facebook, moderators can choose to delete or hide comments, and these arbiters tend to use the latter in order to avoid an escalation of conflicts. The hide function makes comments invisible to participants in the comment section, but the hidden texts remain visible to those who made the comment and their network. Thus, the users are unaware of being moderated. In this paper, we argue that hiding problematic speech without the users’ awareness has serious ramifications for public debates, and we examine the ethical challenges associated with the lack of transparency in comment sections and the way moderation is conducted in Facebook.

Incels, a portmanteau of the term involuntary celibates, operate in online communities to discuss difficulties in attaining sexual relationships. Past reports have found that multiple elements of the incel culture are misogynistic and favorable towards violence. Further, several violent incidents have been linked to this community, which suggests that incel communities may resemble other ideologically motivated extremist groups. The current study employed an inductive qualitative analysis of over 8,000 posts made in two online incel communities to identify the norms, values, and beliefs of these groups from a subcultural perspective. Analyses found that the incel community was structured around five interrelated normative orders: the sexual market, women as naturally evil, legitimizing masculinity, male oppression, and violence. The implications of this analysis for our understanding of extremism and the role of the internet in radicalization to violence are considered in depth.

This paper uses automated text analysis – the process by which unstructured text is extracted, organised and processed into a meaningful format – to develop tools capable of analysing
Islamic State (IS) propaganda at scale. Although we have used a static archive of IS material, the underlying principle is that these techniques can be deployed against content produced by any number of violent extremist movements in real‑time. This study therefore aims to complement work that looks at technology‑driven strategies employed by social media, video‑hosting and file‑sharing platforms to tackle violent extremist content disseminators.