From Laboratories to PLA: The Dangerous Links of Chinese AI and Western Unconsciousness
- Gabriele Iuvinale
- 52 minuti fa
- Tempo di lettura: 10 min
Western companies, including financial giants such as HSBC and Standard Chartered, industrial players such as Saudi Aramco, and even cloud providers such as Amazon Web Services, Microsoft, and Google, are increasingly adopting cheap and open-source Chinese AI models such as Alibaba's DeepSeek and Qwen. This widespread integration, however, poses a serious strategic risk. Such technologies, with documented vulnerabilities and ties to the PLA, serve as a “digital Trojan horse,” fundamentally threatening the national security, digital sovereignty, and democratic values of the West. This is not a commercial advantage, but a devious penetration that demands an urgent and coordinated Western response.
The global artificial intelligence (AI) landscape is undergoing a radical transformation, with China emerging as a dominant force, seriously challenging the technological hegemony of the United States. Chinese AI models like DeepSeek and Alibaba's Qwen, characterized by lower costs and an aggressive open-source strategy, are finding fertile ground not only among major financial and industrial institutions such as HSBC and Saudi Aramco, but also, crucially, among Western cloud giants and large-scale developers, driven by efficiency and cost-effectiveness.
However, this growing and pervasive adoption is not without its shadows. Behind the apparent economic advantage lie systemic risks and profound criticalities for national security, digital sovereignty, and Western democratic values. Governments like that of the United States, and several European nations including Italy, are expressing increasing concerns about the potential exploitation of these technologies for espionage, compromising sensitive data, and imposing authoritarian digital standards. Restrictive measures, such as targeted bans on government use of DeepSeek, testify to an awareness, albeit belated, that the stakes extend far beyond mere commercial competition.
This is not a traditional "AI race," but a total strategic battle fought on multiple fronts: from critical chip hardware to the war for talent, passing through the definition of diverging ethical and regulatory frameworks. Ignoring the warning signs and continuing on a path of uncritical adoption means exposing the very foundations of Western economies and societies to potentially irreversible vulnerabilities.
The Mirage of Economic Advantage: An Invisible Price Tag?
The primary allure of Chinese AI lies in its seemingly negligible cost. Companies like HSBC or Saudi Aramco, drawn by "low-cost" solutions offered by DeepSeek or Alibaba, appear to ignore, or underestimate, the hidden price of this convenience. This isn't just a "bargain" but a deliberately orchestrated strategic ploy by Beijing:
Subsidies as a Weapon: The aggressive pricing of Chinese models doesn't reflect true market costs. Instead, it results from massive state subsidies aimed at saturating the global market, gaining market share, and making the West dependent. Accepting these prices effectively means financing and accelerating the technological rise of a strategic rival, fueling unfair competition that distorts the global AI market.
The "Dependency Trap": By integrating Chinese AI into their operations, Western companies—and, by extension, their economies—create a deep structural vulnerability. In a context of increasing geopolitical tensions, this dependency could be turned into coercive leverage by Beijing, which might disrupt essential services, alter critical functionalities, or impose unacceptable conditions, potentially paralyzing vital sectors of the Western economy.
The False Promise of Open Source: An Open Door for Whom?
China's strategy of making many of its AI models open source is presented as a sign of transparency and openness. In reality, it is an insidious tactic that masks profound risks to security and digital sovereignty:
"Trust Us": A Dangerous Leap of Faith: Relying on open-source models of Chinese origin essentially means taking a leap of faith regarding their integrity and security. Even if the code is theoretically visible, its immense complexity, combined with the vastness and opacity of the training data and the underlying software supply chain (often opaque), makes it nearly impossible to rule out the presence of malicious backdoors, hidden vulnerabilities, or intentional biases. It is crucial to remember that Chinese national security legislation compels any domestic company to cooperate with state intelligence agencies, jeopardizing data confidentiality and integrity.
Propagation of Authoritarian Standards: The widespread adoption of these models risks normalizing and spreading globally digital standards that reflect Beijing's authoritarian values and control approach. This includes conceptions of data management and privacy radically different from Western ones (consider the European GDPR), potential censorship mechanisms embedded in algorithms, and a clear preference for state control over individual liberty. Every line of code, every algorithmic logic, can become a subtle yet effective vehicle for exporting the Chinese digital governance model.
The Dilemma of Western Cloud Giants: The decision by Amazon Web Services, Microsoft, and Google to offer DeepSeek to their customers, while driven by market logic and demand, represents a dangerous weakness and a strategic contradiction. These giants are effectively legitimizing and distributing a technology that their own governments consider a threat. Their argument of "meeting customer demand" is an insufficient excuse in the face of systemic risks to data security, technological sovereignty, and geostrategic loyalty.
The Silent Threat to National Security and Democratic Values: Direct Ties to the PLA
The targeted restrictions on DeepSeek and other Chinese AI models by numerous U.S. and European government entities are not a whim, but a necessary response to a real danger that the West has been slow to fully acknowledge. And the connection with the People's Liberation Army (PLA) makes the picture even more alarming.
The Digital Trojan Horse: Any Chinese AI system integrated into critical infrastructure, corporate networks, or, worse, government systems, can be likened to a digital Trojan Horse. It offers Beijing a potential access point for industrial espionage, the collection of sensitive data (personal, corporate, governmental), and, in crisis scenarios, the possibility of sabotage or manipulation of vital digital operations. Security reports, including those from SecurityScorecard and CIS Center for Internet Security (January-February 2025), have revealed significant vulnerabilities in DeepSeek, such as weak encryption, SQL injection risks, and hardcoded keys, in addition to extremely broad data collection (including typing patterns) stored on servers in China. A publicly exposed DeepSeek database vulnerability briefly revealed millions of records containing chat history, API secrets, and operational data.
Evidence and Clues of DeepSeek's Ties to the PLA: Intelligence sources and recent reports have raised serious and detailed accusations:
A Recorded Future report (June 2025) indicated that the PLA has "most likely rapidly adopted DeepSeek's generative AI models in early 2025" for military intelligence purposes, particularly for OSINT (Open Source Intelligence), based on claims from a Chinese defense contractor who allegedly provided a DeepSeek-based model to the PLA. This suggests a direct and rapid integration of DeepSeek products into Chinese military operations.
ZDNET (March 2025) reported accusations from an anonymous U.S. State Department official that DeepSeek has not only collaborated actively with Chinese military and intelligence sectors but has also attempted to circumvent export restrictions to acquire advanced U.S. semiconductor chips through front companies in Southeast Asia. The same official claimed that DeepSeek regularly shares user data and analysis with Chinese surveillance systems, adding further weight to privacy and espionage concerns.
A South China Morning Post article (March 2025) cited that the PLA is using DeepSeek for non-combat support functions, particularly in military hospitals, to assist in treatment planning and training programs. This provides the PLA with a controlled and secure environment for AI experimentation and integration.
According to a Defense One analysis (February 2025), DeepSeek has been cited over 150 times in procurement records linked to the PLA and associated defense entities. Furthermore, The New York Times reported that dozens of DeepSeek researchers have or have had affiliations with PLA laboratories and the "Seven Sons of National Defence," a group of universities with strong ties to China's defense industry. These affiliations indicate a deep integration between the Chinese private AI sector and the military-state apparatus.
Erosion of Digital Sovereignty: With the adoption of Chinese AI software and hardware, the West risks gradually ceding control over its own critical digital infrastructures and most valuable data. Digital sovereignty is no longer just a matter of where servers are physically located, but of who controls the key technologies, algorithms, and models that process and interpret information, defining power over the flow of knowledge.
The Chip Cold War: An Underestimated Bottleneck: Focus often falls on software, but China's crucial dependence on advanced Western chips (designed by U.S. companies like Nvidia, AMD, and produced by foundries like TSMC in Taiwan, using Dutch ASML machines) is a strategic weapon the United States is (finally) beginning to wield forcefully through stringent export controls. However, this dependence is also a double-edged sword: if China, with its massive investments, were to achieve significant self-sufficiency in this critical sector, the West would lose a fundamental strategic lever and face a truly autonomous and even more formidable technological adversary.
Inevitable Ethical Conflict: The diverging philosophies on AI—state control and stability versus individual rights and freedom—will lead to an inevitable ethical and regulatory conflict. The West must decide whether it is willing to compromise its fundamental principles of privacy, transparency, and freedom of expression for a short-term economic advantage, or if it is ready to defend its values with targeted technological choices.
Future Prospects and Deeper Implications: A Bifurcated Ecosystem?
The complex interplay of these factors outlines future scenarios with profound implications for every sector.
1. Specific Sectoral Impact: Balancing Efficiency and Security
The choice to adopt or reject Chinese AI will have different repercussions depending on the sector and its risk tolerance:
Finance: Banks like HSBC and Standard Chartered, with a strong presence in Asia, might benefit from reduced costs and seamless integration with Asian partners. However, they will have to navigate strict data privacy regulations (like GDPR in Europe) and increasing security concerns from their home countries, especially regarding the management of highly sensitive financial data.
Manufacturing and Logistics: These sectors might be among the quickest to adopt Chinese AI to optimize supply chains, robotics, and production processes, where cost efficiency is a priority and the risk to sensitive data is perceived as lower compared to finance or defense. AI-powered vision systems and automation are particularly appealing.
Healthcare: The adoption of Chinese AI in the Western healthcare sector is likely to face extremely high regulatory barriers due to the ultra-sensitive nature of health data and stringent privacy laws (like HIPAA in the U.S. or GDPR in Europe). Trust and regulatory compliance will be nearly insurmountable obstacles.
Defense and National Security: In these areas, the integration of Chinese AI is almost universally banned or extremely limited for obvious reasons of national security and to prevent potential espionage, manipulation, or sabotage by a strategic adversary.
Public Services and Government Institutions: As already seen with DeepSeek, the trend is towards strict restrictions on all non-domestic AI models, or those from "adversary" countries, due to concerns about surveillance and the security of state data.
2. The "AI Cold War" Scenario: Technological Blocs and Risked Cooperation
The "AI race" is increasingly taking on the contours of a "technological Cold War," a systemic confrontation that extends beyond commercial boundaries:
Formation of Technological Blocs: We are already witnessing the formation of two main technological blocs: one led by the United States (with allies in Europe, Japan, South Korea) focusing on security, supply chain diversification, and standards based on democratic values; the other led by China (with partner countries within the Belt and and Road Initiative or with converging economic interests) offering integrated solutions and lower costs, but with implications for data governance. This will lead to diverging operating systems, connectivity standards, and, crucially, distinct AI ecosystems.
Reduced Interoperability and Global Cooperation: Fragmentation could severely hinder international collaboration on global challenges that require large-scale AI solutions (such as climate change, pandemic preparedness, or humanitarian crisis management), reducing interoperability between systems and networks and slowing down the resolution of universal problems.
Heightened Strategic Competition: Each bloc will seek to attract more nations into its technological sphere of influence through economic incentives, technology-sharing agreements, and the development of standards, turning technology into a primary weapon in geopolitical competition.
3. Future Prospects of Chinese AI: Overcoming Vulnerabilities and Consolidating Position
China is fully aware of its vulnerabilities, particularly concerning chips, and is implementing aggressive strategies to overcome them and consolidate its position:
Acceleration in Chip Self-Sufficiency: Massive state investments and research programs aim to close the gap in the design and production of advanced semiconductors. While the path is arduous, costly, and long, success in this area is vital for China's long-term AI ambitions, and every step forward will reduce Western leverage.
"Unique" Innovation and Large-Scale Data: China might focus on developing AI that excels in specific fields or is optimized for different infrastructures and operating conditions than those in the West, leveraging its immense amount of internal data, often less constrained by strict privacy regulations. This can lead to highly performant solutions in specific sectors.
Global Expansion Through Integrated Agreements: Through initiatives like the "Digital Silk Road" (part of the Belt and Road Initiative), China will continue to push the adoption of its AI and digital technologies in developing countries, offering comprehensive packages (hardware, software, infrastructure, and financing) that the West often cannot or will not match in terms of cost or integration.
Domestic Consolidation: The continued massive adoption of AI within China's vast domestic market will provide an unparalleled "laboratory" and "testing ground" for its models, allowing for rapid iteration and continuous refinement that will further strengthen its capabilities.
Conclusion: It's Time for a Decisive and Coordinated Response
Critical analysis reveals that the current Western approach to Chinese AI is, at best, naïve and fragmented, and at worst, dangerously self-sabotaging. While some nations and agencies act to protect themselves, the lack of a unified and incisive Western strategy leaves wide gaps in the security net. The emerging and increasingly documented ties between companies like DeepSeek and the PLA dramatically reinforce the need for extreme vigilance.
It is imperative that the West awakens from its slumber driven by economic convenience and formulates a cohesive and proactive strategic response:
Absolute Priority to Security: Place national security, data protection, and digital sovereignty above short-term economic advantage, with a clear identification of risks and sensitive sectors.
Massive and Coordinated Investments in Domestic AI: Accelerate investments in AI research, development, and production of AI and chips within democratic countries, not only to maintain technological leadership but to drastically reduce dependence on vulnerable supply chains.
Robust, Value-Based Standardization: Collaborate intensively to establish global ethical and technical standards for AI that reflect democratic values of transparency, privacy, and accountability, actively promoting them internationally as a credible and secure alternative.
Maximum Awareness and Transparency: Educate companies, institutions, and citizens about the real risks associated with using technologies from authoritarian regimes, providing tools for informed assessment and risk mitigation. Information about ties to the PLA must be clearly communicated.
Heightened Geopolitical Coordination: Strengthen existing alliances and forge new collaborations to present a united and coherent front in managing the challenges posed by Chinese AI, avoiding isolated actions that can be easily circumvented.
Ignoring these risks or continuing to prioritize convenience over integrity means allowing a digital Trojan horse to be introduced into our own fortresses, with potentially irreversible consequences for our security, our economy, and, ultimately, our freedom. The stakes are not just technological supremacy, but the very future of our digital order and the values upon which it is founded.
Comments