Questions about the risks posed by artificial intelligence built in China and distributed to Western businesses and governments at little or no cost have drawn growing attention from researchers and policymakers. On a monthly basis, firms such as DeepSeek and Xiaomi release updated iterations of what the industry calls "open-weight" models — a category of software that grants users access to the underlying code without requiring them to pay recurring licence fees.
Procurement decisions across the UK have at times favoured lower-cost technology from Chinese suppliers, with questions about manufacturing conditions and corporate ownership receiving comparatively little scrutiny in those deliberations. Questions about granting Chinese-made electric vehicles open entry to UK dealerships illustrate the broader tension at play. Their lower price points relative to other electric vehicles on the UK market sit alongside unresolved questions about whether state support from Beijing has shaped those costs in ways that place pressure on car manufacturers elsewhere.
Low-cost AI tools have added a new dimension to the risks posed by technology exports. Holding the position of the world's second-largest economy by output, China is entering what its government describes as a new phase of national development. People close to Xi Jinping have been watching, according to reports, for signs that the inherent strains of a highly centralised, command-driven political structure might translate into unrest among the broader population.
A persistent slump in real estate, a mismatch between job openings and the pool of people looking for work, uneven healthcare provision across regions, a skewed ratio of men to women, and a population tilting older — these are among the structural conditions that analysts describe as policy tests the governing party faces in holding onto its position.
Advances in artificial intelligence have expanded the tools available for population monitoring in China, where image-capture technology is now deployed to track the movements and actions of residents. Those who openly challenge the prevailing order face removal from positions of influence, mandatory ideological programmes, or formal penalties. Artificial intelligence has come to occupy a place in the central committee's broader planning framework as a tool of strategic value.
Budgetary pressures tied to domestic needs have not fully eased, yet the country's leadership is now able to direct attention toward external challenges. Expanded trade corridors under the Belt and Road Initiative have given Beijing a platform through which participating nations may find their access to essential goods and infrastructure tied to Chinese suppliers.
Analysts have reported that Russia is deploying artificial intelligence tools to intensify cyber operations targeting European government bodies and commercial enterprises, while China pursues a separate course — broadening the pool of countries that import its products. The option of deploying armed forces remains on the table. Naval expansion by the government in Beijing has added a growing number of vessels to its maritime forces in the Pacific region. Taiwan's position as a leading producer of semiconductors has placed it at the centre of that strategic focus.
Across the AI landscape, teams working out of China have kept budgets notably leaner than those attached to comparable efforts in the United States, all while pushing out new models at a rate that has sometimes hit several releases within a single month. Analysts working in the technology sector have noted that AI models developed in China broaden the overall landscape of the field, offering an alternative to the proprietary systems put out by companies such as OpenAI and Anthropic. Variants tailored for European and Indian markets are understood to be taking shape alongside the existing line.
Meta's open-model development is moving at pace, yet the company — along with other challengers in the field — has yet to build a strong foothold among paying customers. Depending on their priorities, AI users can weigh a lower-cost, adaptable open-source platform against a pricier proprietary package whose full feature set they may not fully utilise.
Burley, who leads the Centre for Information Resilience — a non-profit organisation focused on documenting human rights abuses and countering the spread of disinformation — noted that while properly isolated deployments of DeepSeek's software fall outside the platform's reach, that protection leaves a wider set of concerns unaddressed. Hosting a model on private infrastructure does not, by itself, guarantee that content filtering tied to political considerations has been lifted.
A NIST assessment of downloaded DeepSeek models running on the institute's own infrastructure found politically filtered content built into those versions, indicating that hosting the same models within the UK would leave that filtering intact rather than eliminating it. Burley noted that having such restrictions from Beijing woven into tools that businesses and public services draw on for information and decision-making warranted careful consideration.
According to the terms laid out in DeepSeek's chatbot privacy documentation, user-submitted text and uploaded files are gathered, handled, and retained on servers based in China, with the terms also allowing those records to be handed over to Chinese authorities upon request. According to Michael Wooldridge, an Oxford University professor specialising in artificial intelligence, there are plausible grounds for believing that data entered into the chatbot may be accessible to the Chinese state.
How decisions get made — and by whom — stands to shift in ways that reach well beyond routine adjustments. Debate over the UK's artificial intelligence strategy has centred on whether to reduce reliance on overseas providers — particularly from the United States and China — while directing greater investment toward homegrown and European alternatives. Analysts have raised questions about whether widespread adoption of Chinese-made vehicles and artificial intelligence tools could leave domestic industries hollowed out and expose sensitive data to potential vulnerabilities.
