Quick Buzz Feed
In this commentary, former Italian Undersecretary of Defence and former Chairman of Parliament’s Security Committee Raffaele Volpi argues that artificial intelligence has become a new gateway for strategic influence, making technological choices inseparable from national security. While China poses a growing challenge through AI ecosystems, he contends that Europe’s answer lies not in equidistance between Beijing and Washington, but in building the autonomous capabilities needed to preserve strategic sovereignty.
The participation of the Director General of AgID in the World Artificial Intelligence Conference in Shanghai highlights a significant disconnect in Italy: while the government maintains a clear Euro-Atlantic orientation, public administrations, companies, and research centers often treat technological decisions as separate from national security. Unlike the transparent 2019 Memorandum, Chinese influence now infiltrates through less visible channels such as industrial partnerships, university agreements, cloud services, and inexpensive, easily integrated AI models. Each of these seemingly minor steps gradually builds dependency without requiring high-level political intervention. This fragmentation poses the greatest risk, as Beijing can achieve strategic realignment by accumulating many small decisions—e.g., adopting cheaper software or selecting infrastructure without considering future dependency. By the time the cumulative effect is recognized, the cost of reversing these dependencies becomes substantial.
Artificial intelligence is more than just an enabling technology; it's rapidly becoming the primary conduit through which society acquires information, interprets reality, and forms judgments. Unlike social media, where content is clearly attributed to an identifiable actor, users tend to perceive AI systems as neutral sources, attributing a higher authority to their responses. This perception makes AI an unprecedented intermediary of information, deceptively appearing non-intermediated. However, absolute neutrality in an AI model is a myth, as every system is shaped by its training data, selected sources, instructions, and filters. Without explicit propaganda, subtle manipulations—such as downplaying certain information or marginalizing specific sources—can be employed by those who control these models. This gives controllers an unparalleled tool to shape public opinion and poses a significant "cognitive backdoor" risk.
It's crucial to acknowledge that the challenges posed by AI models are not exclusive to Chinese systems; American models, on which Europe heavily relies, also inherently lack absolute neutrality. The fundamental difference between them lies not in their inherent purity, but in the legal and oversight frameworks within which these technologies operate. An AI model developed in a democracy is subject to legal recourse, parliamentary scrutiny, a free press, and corporate accountability, providing mechanisms for verification and contestation. Conversely, an AI ecosystem embedded within an authoritarian system offers none of these safeguards, as such systems are designed to resist external verification. Therefore, the real distinction for strategic security is an AI system's contestability—its openness to challenge and oversight—rather than an unattainable ideal of pure neutrality.
While the threat from authoritarian AI systems like China's is clear, Europe also faces a distinct, albeit different, challenge in its relationship with the United States. Relying on an ally for critical computing power, cloud infrastructure, and AI models still constitutes a form of dependency. This exposure means that Europe could be subject to regulatory or extraterritorial decisions made by the US for its own national interests, which may not always align with Europe's, as evidenced by past disputes over tariffs and semiconductors. An alliance, by itself, does not guarantee a complete coincidence of interests. Therefore, Europe's strategic response should not be to maintain equidistance between Washington and Beijing—which would be a strategic miscalculation—but rather to develop autonomous European capabilities. This autonomy would transform transatlantic cooperation from a necessity into a deliberate choice, strengthening Europe's position.
To navigate the complexities of AI influence and dependency, a dual approach is essential, focusing on both robust procedures and enhanced capabilities. First, clearer procedural rules are needed for international activities of public administrations in strategic sectors, ensuring better coordination with national security authorities on technological decisions. Companies require advanced technological risk assessment tools that extend beyond mere formal compliance. Universities and research centers need proportionate regulations to differentiate legitimate academic cooperation from the illicit transfer of sensitive knowledge. Second, Europe must significantly invest in its capabilities. Relying solely on prohibitions is insufficient; companies cannot be expected to forgo competitive technology without viable alternatives. Therefore, Europe must strategically invest in computing capacity, cloud infrastructure, semiconductors, large language models, cybersecurity, and dual-use applications. This involves unifying fragmented industrial ecosystems across research, industry, defence, and space to achieve not unrealistic self-sufficiency, but the autonomous capacity to make informed choices.
Maintaining relations with China remains a pragmatic necessity, and complete economic or technological decoupling is an unrealistic goal. However, this cooperation must be balanced with the ability to critically distinguish between standard economic exchange and activities that foster strategic dependency or exert informational influence. The core challenge for Italy and Europe is not merely how effectively China can exploit existing vulnerabilities, but rather the internal capacity to recognize these vulnerabilities before they mature into irreversible dependencies. In the rapidly evolving landscape of artificial intelligence, the concept of national sovereignty has expanded beyond traditional domains like ports, networks, and critical infrastructure. It now fundamentally encompasses the freedom and integrity with which a society forms its own convictions and processes information, underscoring the profound geopolitical stakes of AI development and adoption.