Artificial intelligence may become one of the Chinese Communist Party’s most powerful instruments. It may also become an erosive force against the political order that seeks to control it. These possibilities are not mutually exclusive. AI offers Beijing exceptional capabilities for economic modernisation, administration, surveillance and military power. Yet it also accelerates the production and dissemination of information, can amplify ideological narratives and radicalisation faster than traditional mechanisms of political supervision can respond, and rewards experimentation and informational breadth. In defence, the contradiction becomes sharper: China needs AI to accelerate military modernisation and decision-making, while the People’s Liberation Army remains institutionally founded upon absolute Party control. Together, these pressures raise an old question of communist modernisation in a new form: can transformative productive forces be fully exploited without loosening political authority over the information, institutions and instruments of power through which they operate?
China is emphatically not the late Soviet Union. Its socialist market economy combines a substantial private sector, market incentives, globally competitive companies and extensive international commercial integration with a large state sector and ultimate Party-state authority. Chinese policy itself recognises the private economy as an important component of the socialist market economy and a significant source of innovation. China’s technological record therefore makes the proposition that communist political systems are intrinsically incapable of innovation untenable. Yet the Soviet legacy remains analytically relevant in a narrower sense. It demonstrated how the political requirements of a communist system (ideological discipline, vertical authority and control over information) could eventually interact with the increasingly complex requirements of technological modernisation. China has escaped many of the economic structures that produced Soviet stagnation. Whether it has entirely escaped that deeper institutional tension is another question.
AI makes the question unusually acute because information is simultaneously its raw material and its product. An advanced generative model does not merely reproduce information. It connects, interprets and reformulates it, potentially giving individuals personalised explanations of history, society and politics at enormous speed. This does not mean AI will liberalise China. It could equally strengthen the state through more capable surveillance, censorship and social analysis. But it alters the scale and tempo of information control. Models can generate endless variations of arguments, users can develop indirect means of obtaining information, and political narratives can be personalised rather than publicly broadcast. The relevant issue may therefore become whether Party control can continuously evolve at the same speed as systems whose utility derives partly from generating conclusions that were not individually predetermined.
Beijing’s own regulation illustrates the tension. China’s generative-AI rules encourage innovation, international technological cooperation, improved training resources and indigenous development of algorithms, frameworks, chips and software. Simultaneously, providers must uphold “socialist core values” and prevent outputs involving subversion of state power, overthrow of the socialist system or threats to social stability. Newer regulation of anthropomorphic AI addresses extremism, “historical nihilism,” social-group confrontation and manipulation of users. Radicalisation should not therefore be treated as an established threat to CCP rule, but as a plausible emerging mechanism. AI can intensify nationalism, grievances or ideological convictions in directions that need not correspond either to liberal opposition or to the Party’s preferred narratives. The significant variable could become the widening distance between the speed at which AI generates political meaning and the speed at which authority can define its permissible boundaries.
The political response to this informational challenge could, in turn, affect innovation. China’s great post-Mao achievement was partly to demonstrate that extensive economic opening need not produce equivalent political opening. AI may complicate that separation because information itself has become an increasingly important productive input. Frontier systems benefit from scientific interchange, diverse data, experimentation, intellectual competition and the capacity to discover relationships not prescribed beforehand. Yet the resulting systems must remain politically bounded. China’s present approach captures this duality: Beijing promotes open-source development, international cooperation and rapid AI innovation while simultaneously insisting upon safety, controllability and human control.
Here the Soviet experience offers a warning rather than a template. Soviet technological history was not simply one of failure: the USSR achieved extraordinary results in fields ranging from spaceflight to military technology and fundamental science. The more relevant problem was translating technological achievement into sustained, adaptive innovation across an increasingly information-intensive economy. Ideological rigidity, bureaucratic incentives and restrictions upon information could become friction within the modernisation process. China’s markets, private firms and global economic connections give it mechanisms the Soviet Union largely lacked. The narrower question is whether the political logic inherited from the Leninist tradition can nevertheless impose a marginal innovation cost as AI becomes more general, autonomous and dependent upon complex informational relationships.
Defence and security bring these tensions together. Chinese military modernisation combines “informatization” and “intelligentization” with the continuing requirement of absolute Party leadership over the armed forces. Here AI’s technological logic encounters an exceptionally sensitive political principle. Military AI promises advantage partly by compressing the sequence from sensing to analysis, decision and action. But the PLA is not institutionally conceived as politically autonomous from the ruling party: control of military power is integral to CCP rule. Increasing machine autonomy therefore creates not only technical questions about human oversight but a deeper question about how machine-speed operational effectiveness can coexist with hierarchical political control.
This also gives the Soviet legacy another, limited relevance. Soviet power could sustain formidable military-technological sectors even while weaknesses accumulated elsewhere in its innovation system. Contemporary China has sought something very different: exploiting interaction between civilian technological dynamism and national-security modernisation rather than isolating defence innovation within a largely separate military-industrial sphere. That can be a major advantage. But the increasingly dual-use character of advanced AI also means that political and security restrictions imposed upon frontier technology can potentially travel back into the civilian ecosystem from which future military capability is expected to emerge.
The resulting pressure on Party control is therefore not one-dimensional. Greater AI capability may make information and ideological formation harder to contain; efforts to contain them may impose increasing costs upon the environment producing technological innovation; and strategic competition simultaneously makes restricting AI development increasingly expensive. Defence concentrates these tensions because Beijing requires both machine-speed capability and dependable Party authority over the institutions wielding coercive power.
None of this implies a replay of 1991. China has already demonstrated an adaptability, market dynamism and international economic integration profoundly different from the late Soviet experience. The Soviet legacy matters for a more modest reason: it warns against assuming that technological modernisation and political control necessarily remain complementary indefinitely.
China may continue to reconcile them. But AI subjects that achievement to an unusually demanding test. The more capable the technology becomes, the greater its value to Chinese power; yet the more consequential become the informational freedoms, institutional adaptations and degrees of autonomy associated with that capability. The enduring question is therefore not whether China is becoming the Soviet Union, but whether it has fully escaped an older contradiction of communist modernisation. Artificial intelligence may provide the most demanding stress-test yet.
*Ioannis Sidiropoulos, Legal Consultant, LL.M, MA, PhD (c) AI Governance and Sovereignty
