Beijing has firmly rejected accusations of illicit data collection, asserting its commitment to data privacy amidst growing international scrutiny of Chinese tech firms. During a recent press briefing, China’s Ministry of Foreign Affairs spokesperson, Guo Jiakun, addressed concerns head-on, stating unequivocally that China “has never – and will never – encourage businesses or individuals to collect or store users’ data through illegal means.”
Guo’s remarks were prompted by reports of access restrictions to DeepSeek, a Chinese artificial intelligence service that has rapidly ascended to the top of the US iPhone app store charts. This AI platform’s impressive capabilities, coupled with its remarkably low training expenses, have reportedly sent tremors through the US tech sector, contributing to a noticeable dip in American tech stock values.
The Ministry spokesperson emphasized China’s dedication to data protection, insisting that all operations are conducted within strict legal boundaries. He also took aim at what he described as the “overextension of the concept of national security” and the “politicization” of trade, economic, and technological matters – a thinly veiled critique of Washington’s increasingly cautious stance towards Chinese technology. Simultaneously, Guo reiterated China’s resolve to protect the “legitimate rights and interests” of its companies on the global stage.
DeepSeek, it’s worth noting, has distinguished itself by adopting an unusually open approach to AI development. Its cutting-edge models – including the V3 language model, R1 reasoning model, and Janus Pro vision model – are freely accessible for download. Furthermore, DeepSeek has published detailed research papers outlining their training methodologies, enabling other developers to learn from and potentially replicate their techniques using their own data.
A key feature touted by DeepSeek, and one that differentiates it from proprietary models developed by US giants like OpenAI and Google, is its offline functionality. When downloaded and run locally, DeepSeek’s models operate without an internet connection, theoretically preventing the transmission of user data to third parties. This “privacy-first” design has garnered praise from some corners, including venture capitalist Marc Andreessen, co-founder of Netscape, who hailed DeepSeek-R1 as “a profound gift to the world.” DeepSeek also offers its online chat service without charge, providing global users with tools for tasks ranging from internet meme comprehension to complex problem-solving and idea generation.
However, despite Beijing’s assurances and DeepSeek’s open approach, Washington is unlikely to be entirely convinced. US government officials, speaking on background, have expressed persistent concerns about data security and the potential for Chinese AI platforms to be leveraged for purposes beyond their stated intentions.
“While we acknowledge the statements from the Chinese Ministry of Foreign Affairs, actions speak louder than words,” one US official noted. “The structure of China’s legal and political system raises inherent concerns about data access and control, regardless of stated policies.” This official pointed to China’s National Intelligence Law, which compels organizations and citizens to support and cooperate with state intelligence work, as a potential avenue for government access to data, even if collected by ostensibly private companies.
Another area of concern for the US government is the rapid advancement of Chinese AI capabilities. DeepSeek’s success, while lauded by some in the tech community, is viewed with apprehension in Washington. The fear is not simply about economic competition, but about a potential shift in global AI dominance. “AI is the technology of the future, and control over its development and deployment carries significant geopolitical implications,” another US government source commented. “We need to ensure that the global AI landscape remains open, innovative, and aligned with democratic values, and we are carefully monitoring developments like DeepSeek’s rise.”
Moreover, the US government is likely to scrutinize the “openness” of DeepSeek’s models with a degree of skepticism. While the models themselves may be available for download, questions remain about the data used to train these models and the potential for biases or vulnerabilities to be embedded within them. The lack of transparency around data sourcing in China, coupled with concerns about censorship and state influence, further fuels this skepticism.
Interestingly, DeepSeek’s success appears to be prompting a reaction from its competitors, including US-based OpenAI. Shortly after DeepSeek gained prominence, OpenAI announced that search functionalities for its ChatGPT service would be made freely available, eliminating the need for user sign-ups. Alibaba, a Chinese tech giant, also unveiled its latest open-source AI model, Qwen2.5-Max, trained using methods similar to DeepSeek-V3, and launched a free web service for users to access its models. This suggests that DeepSeek’s challenge is not going unnoticed, and may be spurring a broader industry shift towards more accessible AI models.
Despite the competitive ripples and China’s denials, the US government’s underlying concerns about data security, national security, and the broader implications of Chinese AI advancement are likely to persist. The DeepSeek phenomenon serves as a stark reminder of the intensifying global race for AI supremacy and the complex geopolitical considerations that accompany it. The coming months will likely see continued debate and policy adjustments in Washington as it grapples with the implications of China’s growing AI prowess and seeks to safeguard US interests in this rapidly evolving technological landscape.