Securitizing AI and Cyber Technology is a means of threat construction that reinforces polarization, political violence, and radicalism. Also, turns case because tools used to solve the securitized threat with AI contributes to violence and extremism while spreading misinformation and driving conflict

AI has been increasingly associated with radicalization, extremism, and polarization. Each of these terms has various definitions and have been subject to analysis across a deep and broad literature in many academic dicisplines. There is no consensus view in the literature of what constitutes radicalism or polarization or indeed how they can be measured. There is nothing inherently wrong with taking a radical position on a political, social or economic issue, and some radical beliefs have led to social progress. The term radicalism has also been used to refer to a more progressive and liberal politics [29]. The concern of this article, however, is to look at the connection between radicalism, polarization and violence in the modern context, and the impact of AI on these dynamics. In the context of this paper, radicalism is referred to as a process in which an individual or group adopts beliefs that are outside of the acceptable norms of society including a desire to break with the continuity of social institutions, stemming from a disbelief in their reformability [30]. In the contemporary era, and in popular usage, radicalism stands for political extremism of any variety, of the left or right [31], including such beliefs as white supremacy (the idea that whites are inherently superior to other races), white replacement theory (the notion that white people are being replaced by ethnic minorities), extreme forms of conspiracy theory (those espoused by the QAnon group, for example), and, crucially, the belief that violence against other peoples is acceptable and/or necessary in the pursuance of political goals or motivations. Radical beliefs contribute to polarization and form the basis of it – in this context, political polarization is the process of divergence of political attitudes to ideological extremes [32]. This can entail societies and/or societal groups become divided and through which incompatible and extreme views about the way that society should be governed emerge. Polarization is more likely if large numbers of people adopt radical beliefs, and can be (positively or negatively) influenced by political leaders, including through demonizing opposition politicians or groups, political reforms, curtailing democratic freedoms, and/or pursuing radical social and economic policies [33]. Polarization can also be the context in which radicalization and political violence occurs and/or an enabler for extremist violence [34]. According to recent research, polarization can create a matrix of vulnerabilities that may lead to violent extremism, and there are securitized forms of polarization that have emerged during the period of the post-9/11 war on terror which have contributed to armed conflict and political violence (these will be returned to later in the article).

So what is the connection between AI and these phenomenon? Information and communications technologies are increasingly recognized as having an instrumental effect on radicalization (the process through which an individual adopts radical or extreme beliefs), as well as polarization and the resort to political violence [35]. Technology is rarely the sole factor in these processes, and a mix of socio-economic, cultural, cognitive and historical factors are normally at play. In the post-9/11 period, technology has been instrumentalized by individuals and groups to increase levels of polarization, radicalize and encourage extreme political viewpoints, and in some cases directly encourage and cause political violence. Al Qaeda and ISIS activities online, for example, are aimed at radicalizing people into a belief system in which violence is a core goal, and these groups have contributed to and exploited polarization within societies and within the international system. On the far-right, online groups, sites and networks have also engaged in polarization and radicalization processes, calling, among other things, for white people to be ‘defended’ from the threat of multiculturalism, and creating narratives that depict those with opposing political viewpoints as anarchist, terrorists, or radicals themselves [36]. According to recent research, extremist groups are harrnessing the power of emerging technologies, including the algorithms that exist on social media that divert people's attention to extremist content, through the use of gaming platforms that can be modified to promote or reflect extremist goals or viewpoints, and through the use of AI generative models “that may enable and accelerate the production of a greater quantity and quality of digital propaganda manufactured by non-state extremist actors” [37].

Securitization – a well-established concept stemming from the Copenhagen School of International Relations [38] as well as Foucaultian IR [39] - is integral to these dynamics, although not often recognized as such. Securitization is the process in which political actors use discourse, narratives, and framings to construct or highlight existential threats (terrorism, for example) to a referent object (e.g. the state), which are then used to implement special or emergency political measures [40]. Securitization has already occurred within cyber security debates – the use of threat analogies like ‘digital pearl harbour’, ‘weapons of mass disruption’ and ‘digital 9/11’, as well as biological metaphors such as ‘worms’ ‘viruses’ and ‘infodemic’, are examples of securitizing discourse that has framed cyber security as an existential threat and has arguably led to the centralisation of power by national security agencies [41]. Securitization can contribute to polarization; framing a political adversary as an existential threat is clearly polarizing; arguing, for example, that the Black Lives Matter movement is a threat to America is one such example, as per Whyte's argument above. It can also be used to justify special measures. The threat from terrorism, for example, and the language of the ‘war’ on terror was used to justify mass surveillance and the invasion of Iraq in 2003. The relationship between securitization on the one hand and polarization and radicalization on the other has barely been touched upon in academic debates (despite separate and unconnected literatures on each). But there are clearly some close connections. Securitization can be a process used deliberately to target radical groups for special measures; it can be a polarizing process, particularly because securitizing narratives are politically contested within societies and internationally – the level of division over the use of the term ‘war’ for the fight against Al Qaeda is a case in point; and securitization can lead to excessively militarized responses to political issues that create cycles of conflict and violence in international affairs.

Technology has come to play a crucial role in securitization processes. Social media in particular has amplified securitizing discourse and narratives, allowed a more diverse range of actors to be involved in securitizing processes, and created online spaces where there is no refutation of the types of propaganda, misinformation or disinformation that conjure (often illusory) enemies and threats. As Saedoc and Dezelen argue, social media polarizes because it allows for the synthesis of hate speech and fake news (as well as other dystopian narratives (e.g. sensationalism) with existing prejudices and stereotypes [42]. This divisive discourse contributes to and exacerbates “social fragmentation” in society, especially among groups that are already in conflict because of societal diversity [43]. In the context of the main focus of this article, algorithms may contribute to these dymanics by presenting fringe or radical views as more mainstream, and by diverting peoples’ attention to radical content [44]. As a report by the European Parliament argues, Polarisation can happen by design of technological systems “that potentially narrow the diversity of information accessed by individuals while facilitating the dissemination of divisive and emotionally-charged content” or by deliberate manipulation “through the exploitation of loopholes in an attention-driven media ecosystem to stoke divisions and manipulate users” [45].

It is thus important to consider the multidirectional nature of the AI securitization. It is possible that AI itself has been securitized i.e. it has been framed as presenting an existential threat to states. It could also have played a role during securitization processes relating to other threats – i.e. it has been deployed as part of special measures to confront terrorism, for example. Finally, the securitization of AI may have contributed directly and led to the connected processes of radicalism, polarization and political violence. As this article goes on to argue, the representation of AI as an existential threat, and the political contexts out of which it has emerged, suggest that the securitization of AI has occurred and has already had wide and underappreciated consequences.

3. The rise of algorithmic securitization

The securitization of AI has in fact been a regular feature of contemporary political narratives and discourse. On the one hand, AI is often framed as a tool to be used by national security agencies to respond to radicalism and to disrupt groups or individuals who commit political violence. On the other, the securitization of AI has shaped its use and development and encouraged applications of the technology that have fed into and exacerbated international and societal conflicts, polarization and division.

3.1. History, media and culture in the securitization process

The roots of the securitization of AI predate securitization theory itself. Aside from the longer philosophical debate and references to the possibility of the emergence of Artificial Intelligence, which goes back at least to the 13th Century [46], the reality of machine learning and other related automated processes were kick-started by World War 2. The Turing machine, developed by Professor Alan Turing to break the German enigma encryption codes, was a major breakthrough in modern computing and paved the way for the modern field of AI research to emerge. The Turing machine developed out of a highly securitized war time context in which Britain and its allies faced an existential threat from Germany, a power committed to radical and extreme ideas. In this instance a breakthrough in cryptanalysis and automated reasoning helped defeat extremism, but the fact that progress in the field emerged out of military competition, and was put to use by intelligence agencies should not be discounted in the understanding the broader historical arc of this technology. During the Cold War, it was the intelligence agencies and military establishments which funded many of the advanced Artificial Intelligence research institutes, including the US Defense Advanced Research Projects Agency (DARPA). Later in the Cold War, as recently declassified CIA papers reveal [47], the US Central Intelligence Ageny (CIA) were focused on the predictive power of AI, including its potential utility in the emerging missile defence and space race with the Soviet Union [48]. To this date, centralized and powerful national security establishments around the globe are investing heavily in AI and related technologies for surveillance and war fighting purposes.