WASHINGTON — Beijing’s already strict online censorship is expected to strengthen further after several Chinese cybersecurity firms announced that they had integrated DeepSeek, an artificial intelligence large language model, into their services.
DeepSeek, an AI startup based in China’s Zhejiang province, launched the open-source language model DeepSeek-R1 in January. The model has received praise from Chinese officials and impressed global AI experts with its performance, which rivals the most advanced Western models at a significantly lower cost.
TopSec Technologies Group, one of China’s oldest cybersecurity companies, announced this month that it had incorporated DeepSeek into its AI model “Tianwen” that would “enhance its existing security products and services.”
QAX, a major player in China’s cybersecurity sector, also announced a full integration with DeepSeek, claiming it had “significantly” improved the company’s threat assessment capabilities.
NetEase’s cybersecurity unit, Yidun, said DeepSeek could efficiently monitor online content and “precisely identify hidden risks” in text, such as homophones, character splitting, and metaphors — common tactics used by Chinese internet users to evade censorship.
Many of these firms integrating DeepSeek provide internet monitoring services to both private and state-owned enterprises, as well as China’s central and local governments, intensifying concerns over the country’s already stringent censorship.
Some of the companies even have close ties with the authorities. For instance, TopSec Chairwoman Li Xueying is a deputy to the Beijing Municipal People’s Congress, and QAX Chairman Qi Xiangdong is a member of China’s top political advisory body, the Chinese People’s Political Consultative Conference.
More online censorship and harassment
Due to their advanced language processing capabilities, AI-powered tools like DeepSeek could dramatically enhance internet censorship.
“The government and its contractors can use AI tools like DeepSeek to sift through and categorize vast amounts of data, analyzing trends in political, military, and social issues,” Eugenio Benincasa, a senior researcher at the Center for Security Studies of ETH Zurich, told Voice of America (VOA).
“These tools can, for example, quickly flag potential sources of instability or shifts in public sentiment, giving the authorities more time to respond in ways that align with their objectives,” he added.
AI models can also be used to facilitate China’s cyber influence operations. Last week, U.S. AI company OpenAI released a report stating that suspected China-based accounts had used ChatGPT to generate English-language comments attacking Chinese dissident Cai Xia on social media.
Leader in surveillance technologies
China has been a global leader in AI-driven surveillance, particularly in facial recognition and biometric data collection. These technologies have been widely deployed in regions like Xinjiang and Tibet to monitor residents’ daily lives. During the COVID-19 pandemic, similar surveillance tools were used nationwide as part of Beijing’s containment measures.
Major AI-powered monitoring projects in China include the “Smart City” initiative, “Operation Sky Net” and the “Sharp Eyes Project” launched in 2015, which enable extensive public surveillance by deploying security cameras on a massive scale in public spaces.
China also exports its surveillance technologies to other countries, particularly “swing states” — countries that are somewhere between between democracies and autocracies.
Chinese firms Hikvision and Dahua Technology, for example, jointly make up roughly 34% of the global market for surveillance cameras.
China’s push for AI self-reliance
A month after DeepSeek launched its language model, major Chinese tech companies such as Tencent, Baidu and Alibaba — along with banks, hospitals, and various government agencies — rushed to integrate the technology into their systems.
The rapid adoption of DeepSeek aligns with Beijing’s long-standing push for technological self-sufficiency, analysts said.
Dakota Cary, a China-focused expert at cybersecurity firm SentinelOne, told VOA that while some of these announcements may be motivated by marketing and publicity purposes, DeepSeek’s impact is real.
“What’s clear is that DeepSeek’s architectural improvements have made AI models more efficient, and this is already influencing global AI research,” he said.
The Chinese government has explicitly prioritized the research and development of AI as a key pillar of its national technology strategy. In 2018, Chinese President Xi Jinping stated that “accelerating AI development is a strategic issue to decide whether we can grasp opportunities.”
Steven Feldstein, a digital surveillance expert at the Carnegie Endowment for International Peace, told VOA that Beijing has long viewed its reliance on Western technology as a major weakness, and China’s decoupling from the West in the high-tech sector will continue.
“You’ll see China doubling down on efforts to build a self-sufficient tech ecosystem,” he said.
To read the original story in Chinese, click here.
