To work together or warn against? That is the AI question

Deepening threat

The global proliferation of AI is rapidly evolving into a complex security and geopolitical challenge, one no less perilous than the nuclear arms race of the 20th century. Celebrated for their efficiency and convenience, smart devices or algorithm-driven applications can now become strategic vulnerabilities. The paradox is stark: the very technologies designed to simplify life could be weaponised at critical moments.

The global semiconductor crisis that unfolded between 2020-23 laid bare this reality, extending well beyond production delays and exposing the fragility of the entire international economic system. Vital industries—from automotive to medical and defence—were revealed to be overwhelmingly dependent on a tiny group of suppliers based mainly in Taiwan and South Korea. When semiconductor plants in East Asia shut down during the COVID-19 pandemic, millions of vehicles went unassembled, and hospitals faced shortages of essential care equipment.

The danger became evident: interdependence, once a hallmark of globalisation's strength, can swiftly become a choke point if a country chooses to restrict exports or impose sudden controls. In response, the US tightened its export regulations on advanced chips to China and, according to Reuters reports in August 2025, even embedded covert tracking devices in sensitive shipments.

Supply chains—once symbols of openness—are now strategic flashpoints, fraught with suspicion and surveillance. The threat is more acute in the realm of cyber-physical systems that underpin daily life. The modern smart car is as much a vehicle as it is a mobile computer equipped with dozens of chips and internet-connected sensors.

A study conducted by RunSafe Security in August 2025 found that more than 75% of American drivers lack confidence in the safety of their connected vehicles, fearing the prospect of remote hacking. A hostile actor who gains control over navigation or braking systems could disable thousands of vehicles simultaneously, triggering mass traffic disruption or even fatal collisions. Threat analysts think this could be timed to coincide with a national election or global sporting event, paralysing entire cities, destabilising the economy, and severely eroding public trust in government.

Layers of complexity

Smart cities are even more vulnerable. Their infrastructure runs on tightly interconnected networks managing electricity, water, transport, and healthcare. A coordinated cyberattack on a power grid could plunge a metropolis into darkness, shut traffic and transport systems, turn the pipes off, and end hospital operations, making civilian infrastructure a military target and a powerful instrument of political coercion.

A 2025 study by researcher Brooke Kidmoz at the Technical University of Denmark describes this "broad threat landscape" in which cyberattacks can strike at the vital arteries of society—from emergency services to energy networks—making cybersecurity a matter of daily survival, not just data protection.

The stakes are further amplified as data has become the most strategic resource of the 21st century, the digital equivalent of oil, defining geopolitical leverage. To this end, in April 2025 China enacted new legislation requiring foreign companies to store sensitive data within its borders and to submit it to stringent security reviews before any transfer.

This shows how data now constitutes the frontline in defending national sovereignty, accelerating a global race to construct 'digital walls'. This fragments cyberspace into closed zones of sovereignty, undermining digital globalisation and transforming the once free flow of information into a contested, strategic battleground.

Military dimension

The risks have reached unprecedented heights in the military domain. In September 2025, The Wall Street Journal revealed that Ukraine had deployed swarms of AI-powered drones capable of making collective decisions in real time without direct human oversight. If true, this represents an historic precedent, AI no longer supporting armed actors but become its own autonomous battlefield operator.

This throws up ethical and legal questions, such as who is accountable if these drones commit a catastrophic error by striking civilian targets, and how stability can be preserved when the speed of algorithmic decision-making surpasses human reaction. Any new arms race could now see control of warfare become increasingly difficult, with conflicts escalating beyond the bounds of human restraint.

A further question then arises as to the extent to which one tech power's automation will be allowed to connect to those of others. Would, for instance, the US ever allow Chinese autonomous vehicles to operate on its roads, directly connected to American traffic and emergency networks?

For now, that appears unlikely, given that Washington treats any Chinese technology as high-risk in sensitive sectors such as telecoms, energy, infrastructure, and cloud computing. In recent years, the US has pressured allies to exclude Huawei from Europe's 5G networks, showing how commerce can be tied to national security.

Mitigation measures

The reality of the future of AI is likely to fall somewhere between total prohibition and full acceptance. Foreign systems may indeed operate but within isolated networks, shielded from sensitive government data. Likewise, foreign firms may be required to run their software locally under strict oversight, or submit to rigorous security testing before being commercially deployed.

In truth, there can be no absolute guarantees to prevent these technologies from hiding hidden vulnerabilities or 'backdoors,' there can only be risk mitigation through a comprehensive framework including risk assessments, pre-market auditing mechanisms, and classification systems that define security levels.

The European Union has set a precedent in this area through its Artificial Intelligence Act, due for phased implementation by 2027. Meanwhile, the US and EU have tightened export controls on advanced semiconductors to prevent their transfer to strategic rivals, while China has introduced laws requiring foreign firms to store sensitive data domestically and submit to security assessments before any cross-border transfers.

These measures reflect a broader contest over digital sovereignty. The idea of globally harmonised smart cities now clashes with the politics of a sharp technological East-West divide, each bloc with its own standards for chips, communications systems, and cloud infrastructure. Nevertheless, in less sensitive sectors such as consumer applications, space for cooperation remains—provided local laws and regulatory frameworks are respected.

Compliance and regulation

In this context, Europe may emerge as a regulatory power, shaping global standards while maintaining conditional engagement with both sides. The notion that East-West trade will be reduced to traditional goods is overstated. Trade is more likely to be restructured than reduced. Sensitive technologies will be ring-fenced under strict controls, while lower-risk goods and services will continue to flow.

This enhances the importance of intermediary countries such as Mexico and Vietnam, which are becoming alternative hubs for assembly, logistics, and re-export. Technological bifurcation will come at a cost: higher compliance expenses due to multiple certification processes and security protocols, slower cross-border innovation, and increased public investment in 'technological autonomy' initiatives—such as developing domestic semiconductor production or creating sovereign AI models.

These shifts will also open new avenues, such as growth in auditing and cybersecurity services, the emergence of dual-market products (one version for Western markets, another for Eastern markets), and the proliferation of locally-built AI systems aimed at reducing external dependency.

The autonomous vehicle offers a case in point. In the West, components must originate from trusted suppliers, data must be stored and processed locally, and over-the-air updates are tightly regulated. In the East, similar restrictions apply to Western components, with mandatory data localisation requirements. Cooperation between the two spheres will remain minimal, confined largely to non-sensitive data such as weather and mapping information, and facilitated through "trust gateways" designed to prevent raw, sensitive data from crossing borders.

New map of influence

AI is no longer merely a tool for enhancing everyday life—it has become a central axis of political, economic, and security competition among global powers. From the semiconductor supply crisis to vulnerabilities in smart vehicles and cities, from the battle over data sovereignty to AI's role in warfare, a new global order is emerging—one in which technological boundaries increasingly define spheres of influence.

It remains to be seen whether deep interconnection will compel reluctant cooperation between rival powers, or whether we are moving towards a technological schism that divides the world into two distinct camps. What is certain is that AI will remain at the heart of this equation—shaping the future of the global economy and redrawing the geopolitical map.