Britain Must Resist the Appeal of Inexpensive and Risky Chinese AI Solutions

The rise of artificial intelligence (AI) systems developed in China poses significant challenges for Western nations, as companies and governments are increasingly offered these technologies at minimal costs. Notable Chinese developers like DeepSeek and Xiaomi regularly release advanced models that allow users to leverage core software without incurring hefty licensing fees.

In the UK, the popularity of affordable Chinese technology often overshadows concerns about its origins and ethical implications. The debate surrounding the accessibility of Chinese electric vehicles in British retail showcases this phenomenon. Many consumers are drawn to the lower prices while overlooking the potential risks associated with subsidies from the Chinese government that aim to bolster exports and challenge domestic automakers.

The implications of inexpensive AI technology could be even more profound. With China continuously advancing its economic capabilities, the nation is strategically positioning itself for a resurgence on the global stage.

A decade ago, officials close to Xi Jinping were wary of the internal vulnerabilities inherent in a totalitarian state, which could ignite social unrest, reminiscent of events in Tiananmen Square. Today, challenges like a sluggish property market, high unemployment, inadequate health services, and an ageing demographics landscape could potentially destabilize the ruling Communist Party.

However, the advent of AI has granted the Chinese government unprecedented surveillance capabilities. This technology enables the state to track its citizens and monitor their activities closely. While the general populace retains some freedoms to engage in daily activities, such as public gatherings, those considered dissenters face severe consequences, ranging from re-education to punishment.

For the central government, AI is a twofold blessing. It alleviates some financial pressures domestically while allowing leaders to focus on external threats. Unlike Russia, which uses AI to intensify cyberattacks on European entities, China seeks to foster friendly relations with other nations to promote its goods, often without raising uncomfortable topics.

Additionally, the Chinese government is contemplating military expansion, redirecting funds previously used for domestic subsidies towards enhancing its naval capabilities, echoing pre-World War II Japan. Taiwan, a critical player in the semiconductor industry, is particularly of interest.

Chinese AI developers are rapidly releasing new models each month at a fraction of the cost compared to American counterparts. Some analysts assert that these models contribute to a diverse AI ecosystem, complementing the proprietary systems developed by firms like OpenAI and Anthropic.

Conversely, consumers are faced with a choice between cost-effective Chinese AI solutions and pricier American software that includes a range of additional features, often unnecessary for typical users. Yet, there are inherent risks with the Chinese options. As noted by Ross Burley, director of the Centre for Information Resilience, relying on these models can embed political censorship and security vulnerabilities within business and public sector systems.

For instance, an evaluation from the U.S. National Institute of Standards and Technology revealed evidence of political biases in DeepSeek models, indicating that hosting these models outside China would not eliminate their underlying censorship issues.

As stated by academic experts, including Michael Wooldridge from the University of Oxford, there is a reasonable concern that data entered into these AI systems could be accessible to the Chinese government. The ramifications for policymaking are significant; the UK must carefully assess the allure of inexpensive Chinese AI and instead focus on regulating American solutions while nurturing domestic and EU technological development.

Dependence on Chinese technology, whether in vehicles or AI, poses risks of eroding local industries and threatening data security.