China wants to give its AI ’socialist values’ – is it possible?
When it comes to the hidden plumbing of artificial intelligence, Chinese cybersecurity experts are increasingly focused on building a controllable digital core.
Developing domestic AI models that “embody national sovereignty, inject socialist core values and integrate professional expertise in safety and security” is essential for China, according to Wu Shizhong, an academician at the Chinese Academy of Engineering writing in the journal China Information Security in August.
“By building an internally deeply controllable and externally strictly defended cognitive security core, [we will] achieve a systemic leap in national cyberspace defence capabilities,” the network and information security technology expert said.
So how might this be achieved? Wu made several suggestions, such as establishing an identity, security authentication and audit mechanism for AI agents, with their behaviour logs maintained in such a way that agents are registered and can be audited.
Wu also noted that the scope of AI safety is expanding to “national security, digital infrastructure and public trust” – in other words, going far beyond preventing attacks on models.
Rather, he said, it “extends to macro-level issues, such as the lowering of barriers to biological or chemical weapons design and the potential collapse of systemic trust within society.”
Wu also addressed the issue of AI sovereignty, saying sovereign AI is a core concept in great power competitions.
“Other major powers, proactively responding to the US technological monopoly and its trend towards the quasi-militarisation of technology, have abandoned earlier illusions of technology without borders.
“Instead, they are now prioritising full-stack control over their own AI technology, strengthening their sense of technological sovereignty, drawing red lines for AI confidentiality and reinforcing cybersecurity frameworks,” he said.
While Wu’s analysis was focused on back-end infrastructure and cyber resilience, Beijing has been pursuing front-end ideological alignment in public-facing AI applications.
Last year, President Xi Jinping called for efforts to develop new information technologies and apply them to improve oversight of cyberspace and promote socialist values.
This is part of China’s broader efforts to use new technology, especially AI, to guard against Western influence and reinforce its own ideology.
In 2023, the Chinese government mandated that AI-generated content adhere to “socialist core values”. To that end, it rolled out principles for public ethics and individual morality aimed at fostering a modern, harmonious and prosperous society under party leadership.
Despite strict ideological controls, open-source AI models developed in China are gaining traction across American tech hubs.
During recent chats with AI start-up founders in New York, some entrepreneurs told me about their enthusiasm for Chinese open-weight models due to their performance, efficiency and accessibility.
At the same time, others working on AI regulations are not so optimistic. They have raised flags urging caution against hasty development without guardrails.
As AI evolves from a digital tool into a structural pillar of society, the question of whose values are baked into its core becomes inescapable.
Whether we view it through the lens of sovereign digital defence or open-source global collaboration, one thing is certain: the rules of the AI era are being written right now.
Best,
Holly
| 
_20260904162845.jpg)
