Securitising artificial intelligence
Sovereignty, political discourse and the legitimation of algorithmic power in Europe and the United States (2020–2025). Between 2020 and 2025, artificial intelligence moved from
Sovereignty, political discourse and the legitimation of algorithmic power in Europe and the United States (2020–2025)
Between 2020 and 2025, artificial intelligence moved from the status of a promising technological innovation to that of a central issue of sovereignty, security and power. The adoption of the AI Act in Europe, the proliferation of American Executive Orders, the publication of the NIST AI Risk Management Framework and of the Blueprint for an AI Bill of Rights all bear witness to an unprecedented acceleration in the politicisation of AI.
But this shift is not merely a regulatory dynamic. It is a deeper displacement: artificial intelligence becomes a structuring discursive object through which democracies redefine their relationship to power, to risk and to legitimacy.
The question is not simply how to regulate AI. The question is how the discourse on AI transforms the very nature of political authority.
AI as a marker of a crisis of sovereignty
Contemporary states face a structural tension. On the one hand, computational capacities, cloud infrastructures, foundational models and data flows are mostly controlled by transnational private actors. On the other, citizens’ expectations regarding protection, security and democratic accountability remain directed at public institutions.
This asymmetry creates a fragility. States can neither abandon the regulation of AI nor claim to fully control its technical levers. Discourse then becomes a strategic instrument. By speaking of “strategic autonomy”, “digital sovereignty” or “technological leadership”, governments symbolically reconstruct a capacity for mastery.
Artificial intelligence thus acts as a marker of a latent crisis of sovereignty. It exposes dependence on digital supply chains, on semiconductors, on data centres and on global platforms. It highlights the fact that algorithmic power is distributed unevenly, often outside the direct reach of state control.
Faced with this, political language is not incidental. It is constitutive. It redefines what it means to govern in a complex technological environment.
Securitisation: turning a technical object into an existential issue
Securitisation theory helps to analyse this process. An object becomes a security issue when it is presented as a threat serious enough to justify exceptional measures. Security is not given; it is produced by discourse.
Between 2020 and 2025, AI was progressively associated with a series of systemic risks: information manipulation, discriminatory bias, cyber vulnerabilities, strategic dependence, the automation of critical decisions, loss of control over autonomous systems.
This framing produces several effects.
First, it broadens the legitimate scope of public intervention. AI is no longer one industrial sector among others; it becomes critical infrastructure.
Second, it justifies reinforced inter-institutional coordination. In Europe, the creation of dedicated structures such as the AI Office is part of this logic. In the United States, the articulation between the White House, NIST, the OSTP and the federal agencies reflects a strategic centralisation.
Third, it normalises the idea that the governance of AI belongs to the register of national security or of democratic survival. Once this threshold is crossed, the debate changes in nature. It is no longer only about economic efficiency or innovation, but about existential protection.
Securitisation here functions as a mechanism of legitimation. It authorises political choices by placing them within a horizon of necessity.
The European model: normative securitisation
In Europe, the dominant discourse associates AI with trust, with fundamental rights and with the protection of democratic values. The AI Act rests on a risk-tiered approach, structured around the idea that certain applications must be strictly framed, or even banned.
This legal architecture reflects a specific political rationality. Legitimacy does not come from technological dominance, but from the capacity to frame innovation according to normative principles. Digital sovereignty is conceived as the capacity to impose standards, to define the rules of the game, to export a regulatory model.
This strategy extends a pattern already observable with the GDPR: turning regulation into an instrument of global influence. The European Union, lacking technological giants comparable to the major American companies, invests in the norm as a lever of power.
The European discourse insists on the prevention of risks, algorithmic transparency, the responsibility of providers and the protection of citizens. AI is securitised in the name of the defence of rights and of the preservation of a digital space aligned with European values.
Sovereignty here takes a legal and moral form. It is less associated with competition than with the capacity to impose a framework of compliance.
The American model: strategic securitisation
In the United States, the discursive register differs appreciably. Artificial intelligence is integrated into a logic of global competition. It is presented as decisive for national security, military superiority, economic growth and the maintenance of technological leadership against China.
The Executive Orders insist on the need to develop AI safely and responsibly, but always within a framework that presupposes the centrality of American power. The NIST AI Risk Management Framework aims to frame risks, but without stifling innovation.
Regulation is conceived as a tool of strategic optimisation, not as a normative end in itself. Legitimacy rests on the capacity to stay ahead in the technological race.
The discourse mobilises categories such as “national security”, “critical infrastructure”, “global leadership”. AI is securitised as a strategic resource. It becomes an element of the systemic rivalry between great powers.
Sovereignty, in this framework, is expressed through mastery of innovation chains, protection of industrial capacities and consolidation of technological alliances.
Two rationalities, one same objective
Despite their differences, the two approaches pursue a common objective: to preserve the centrality of political power in an environment dominated by complex algorithmic architectures.
In Europe, this centrality operates through the norm.
In the United States, it operates through power.
In both cases, the securitisation of AI makes it possible to reconfigure the relationship between the state and private actors. Technology companies are no longer merely economic partners; they become strategic actors, integrated into hybrid governance arrangements.
Algorithmic power is not abolished. It is integrated, framed, sometimes instrumentalised. Political discourse serves as mediation between technological autonomy and democratic legitimacy.
Governmentality and algorithmic power
Beyond securitisation, another dynamic is at work: that of a discursive governmentality. Institutions produce narratives of mastery, control and anticipation. They organise the perception of risk, define responsibilities and steer conduct.
Speaking of “trustworthy AI” or “responsible AI” is not neutral. These categories structure expectations, configure industrial practices and establish implicit norms of behaviour.
Regulation becomes a language of government. It is not limited to prohibiting or authorising; it shapes a horizon of compliance.
Artificial intelligence, in this framework, does not merely optimise processes. It modifies the very modalities of public decision-making: greater recourse to predictive analysis, partial automation of risk assessments, delegation of certain functions to computational systems.
The question is no longer only that of controlling AI. It becomes that of the coexistence between political power and algorithmic power.
The geopolitical implications
The securitisation of AI does not unfold in a strategic vacuum. It is set within a context of heightened rivalry with China, of tensions over semiconductors, of a reconfiguration of technological value chains and of a growing fragmentation of the global digital space.
The discourse on technological sovereignty contributes to this fragmentation. It legitimises policies of strategic autonomy, export restrictions, selective technological alliances.
AI thus becomes a pivot of geopolitical recomposition. It structures coalitions, redefines dependencies and reinforces the strategic dimension of the digital sphere.
The boundary between internal regulation and external strategy blurs. Governing AI at home becomes an act of positioning abroad.
Conclusion: power in the algorithmic age
Artificial intelligence is not only a field of innovation. It is a space of symbolic struggle for the definition of sovereignty and of democratic legitimacy.
By securitising it, Western democracies do not merely frame a technology. They attempt to maintain their authority in a world where the capacity to compute, predict and optimise becomes an autonomous source of power.
The real stake is not whether AI will be regulated. It will be. The stake is who defines the discursive framework within which this regulation takes on meaning.
The European norm and American power represent two distinct responses to one and the same historical transformation: the emergence of an algorithmic power capable of influencing decision, perception and social organisation.
To govern AI, today, is to govern the way the political sphere accepts, frames or integrates this new centre of gravity of power.
Bibliography
Amoore, L. (2020). Cloud Ethics: Algorithms and the Attributes of Ourselves and Others. Duke University Press.
Aradau, C., & Van Munster, R. (2007). Governing terrorism through risk: Taking precautions, (un)knowing the future. European Journal of International Relations, 13(1), 89–115.
Balzacq, T. (2011). Securitization Theory: How Security Problems Emerge and Dissolve. Routledge.
Bigo, D. (2002). Security and immigration: Toward a critique of the governmentality of unease. Alternatives: Global, Local, Political, 27(1), 63–92.
Dean, M. (2010). Governmentality: Power and Rule in Modern Society. SAGE.
DeNardis, L. (2020). The Internet in Everything: Freedom and Security in a World with No Off Switch. Yale University Press.
European Commission. (2020). White Paper on Artificial Intelligence – A European Approach to Excellence and Trust (COM(2020) 65 final). Brussels.
European Commission. (2021). Proposal for a Regulation Laying Down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act) (COM(2021) 206 final). Brussels.
Foucault, M. (2004). Naissance de la biopolitique: Cours au Collège de France (1978–1979). Gallimard/Seuil.
Hansen, L. (2006). Security as Practice: Discourse Analysis and the Bosnian War. Routledge.
Huysmans, J. (2006). The Politics of Insecurity: Fear, Migration and Asylum in the EU. Routledge.
Krasner, S. D. (1999). Sovereignty: Organized Hypocrisy. Princeton University Press.
National Institute of Standards and Technology (NIST). (2023). Artificial Intelligence Risk Management Framework (AI RMF 1.0). U.S. Department of Commerce.
Nye, J. S. (2010). Cyber Power. Harvard Kennedy School.
Office of Science and Technology Policy (OSTP). (2022). Blueprint for an AI Bill of Rights. The White House.
Rose, N. (1999). Powers of Freedom: Reframing Political Thought. Cambridge University Press.
Rouvroy, A., & Berns, T. (2013). Gouvernementalité algorithmique et perspectives d’émancipation. Réseaux, 177(1), 163–196.
The White House. (2023). Executive Order on Safe, Secure, and Trustworthy Artificial Intelligence. Washington, DC.
U.S. Congress. (2020). National Artificial Intelligence Initiative Act. Washington, DC: Government Publishing Office.
Zürn, M. (2018). A Theory of Global Governance: Authority, Legitimacy, and Contestation. Oxford University Press.