The United States may soon follow Australia in banning China's DeepSeek artificial intelligence from government devices. US Representatives Josh Gottheimer and Darin LaHood have introduced a bipartisan bill to prohibit its use, citing national security risks.
In a letter to 47 US governors and the mayor of Washington, the lawmakers warned that DeepSeek could threaten government cybersecurity, compromise sensitive data, and infringe on Americans' privacy. NBC News reported on March 3 that the proposal follows growing concerns about China's use of AI in mass surveillance and cyber operations.
While China denies these allegations, experts argue the concerns are well-founded. The Chinese government has reportedly leveraged AI models for extensive surveillance, including biometric data collection and social media monitoring for security agencies and the military. Chinese dissidents abroad have also been targeted through information operations.
DeepSeek is already being integrated into services provided by TopSec, QAX, and NetEase—three major Chinese cybersecurity firms. TopSec and QAX directly serve the Chinese government, while NetEase has stated that DeepSeek will enhance its censorship and surveillance tools.
China has spent the past decade investing in AI-driven surveillance, including biometric data capture, facial recognition, and urban monitoring programs like the Skynet project and the Sharp Eyes initiative, which track individuals' movements in public spaces.
In January, Canadian cybersecurity firm Feroot Security discovered embedded code in DeepSeek's login processes that share user data with China Mobile, a state-owned telecom company. The Associated Press described the code as a ''heavily obfuscated'' script that connects to China Mobile's infrastructure. The US previously banned China Mobile in 2019 over national security concerns.
Additionally, Chinese cyber actors have reportedly used ChatGPT alongside DeepSeek to generate phishing emails and disinformation targeting the US, OpenAI disclosed in its February report. OpenAI identified and blocked accounts originating from China that were engaged in influence operations, such as Qianyue Overseas Public Opinion AI Assistant, which analyzed online discussions on Chinese politics and human rights and relayed insights to Chinese authorities.
A network of banned ChatGPT accounts was also linked to influence campaigns in Latin America and Spain, producing articles critical of US policies. These narratives were distributed through national media outlets, aligning with Beijing's messaging. Some of the articles, reviewed by VOA, accused the US of hypocrisy in imposing sanctions on Iran, criticized support for Ukraine as fueling war escalation, and portrayed American economic policies as damaging global stability.
Three articles published in Peru's Wapa, El Popular, and El Plural focused on social issues in the US, depicting homelessness, child nutrition, and crime as severe crises. One piece falsely claimed that most American children experience hunger on weekends and holidays due to government neglect.
While these articles covered a range of issues from human rights to foreign policy, they consistently framed the US as a nation in decline, reflecting a common narrative promoted by Beijing.
With mounting evidence of China's use of AI for surveillance and influence operations, US officials continue to push for stricter regulations to counter potential threats posed by DeepSeek and other AI models tied to the Chinese government.
