Recent weeks have seen a mounting chorus of praise from Chinese companies and other organizations for the new national AI champion DeepSeek. At Rest of World, Kinling Lo noted its integration in cars, smartphones, household appliances, and healthcare, as well as government departments. Other recent reports have noted signs of the adoption of AI technologies in surveillance, censorship, and even non-combat military applications. A recent post at Diyin, from which an excerpt is translated below, compiled more than a dozen local governments’ declarations of DeepSeek’s benefits for monitoring and managing situations online and off:
According to officials in Beijing’s Changping district, the City Management Command Center has used Deepseek to break down “information barriers in grid management,” accurately meeting complex needs like “cross-domain collaboration.” In addition, the local government has integrated DeepSeek’s deep analysis technology with HD video from the Sharp Eyes project (a rural grid management-based “mass public security prevention and control project” using networked video surveillance), establishing an “All-Weather Urban Awareness Network,” which has “eliminated the need for time-consuming, labor-intensive patrols.”
The Public Security Bureau in Inner Mongolia’s Uxin Banner said that DeepSeek has improved the precision of safety and security work for major events and proven effective at preventing and controlling potential security risks. They explained that DeepSeek’s real-time analysis of factors such as participant information and event-site conditions can promptly detect unusual activity and provide early warning, “allowing for flawless safety and security work.”
The Public Security Bureau in Chongqing’s Rongchang district said that since DeepSeek went online, analytical law-enforcement tasks that would previously have taken three officers three days can now be accomplished by a single officer in 15 minutes, with “outstanding results in the field.”
A community propaganda worker in Shenzhen’s Bao’an district said that DeepSeek allows them to handle “public opinion” more “skillfully,” and has greatly boosted their productivity by helping them to quickly grasp “key public-opinion topics” and perform “quantitative analysis of public-opinion trends.” The head of the district’s neighborhood law-enforcement team also indicated that DeepSeek enabled them to more accurately pinpoint the sources influencing public opinion, anticipate risks, and provide strong support for their work responding to public opinion.
Cyberspace Administration authorities for Inner Mongolia’s Hinggan and Xilingol Leagues noted DeepSeek’s marked advantages in recognizing contextual complexities and identifying potential hazards, and said they would continue to push ahead with the use of AI in areas such as content management, “public opinion analysis,” and cybersecurity.
Shandong Province Internet Media Group, a province-level Party media organization, offered more details on DeepSeek’s application in “public opinion monitoring.” The group indicated that since it adopted DeepSeek, it has seen great advances in both its efficiency at parsing “public opinion data” from across the internet and its ability to filter out noise, and that it can now more quickly identify potential hazards when monitoring hot topics. In addition, DeepSeek can automatically generate “strategic recommendations for public opinion response” based on its analysis of vast quantities of data, and provide smarter “suggestions for handling public opinion.”
Topsec, an internet security monitoring company that works closely with the authorities, said that it is already using DeepSeek’s deep content recognition technology—together with other methods such as keyword scanning and optical character recognition—to comprehensively monitor sensitive information and promptly block activity that violates regulations.
Many local propaganda departments and Party media outlets have said that DeepSeek can automatically generate press releases based on real-time information, helping state-media journalists to quickly compose news reports.
The municipal government of Xinxiang, Henan, released a guide to a Smart Official Document Composition Assistant powered by DeepSeek. This reportedly includes a corpus of “official document templates from Party and state administrative units,” and can automatically cross-reference with the latest official lists of words to be avoided or treated with caution in public materials. It can also automatically screen for core political terminology like the “Two Safeguards” and “Two Establishes” and compare it against central [government] documents.
The Party committee for Altay in Xinjiang said that DeepSeek enables thoroughly “smart” Party-building work. If you want to study central policy documents and really grasp their spirit, you can simply upload them and DeepSeek will generate a summary of the key points, highlighting and explaining specialist jargon and implementation challenges (e.g. “grassroots Party organization election procedures”), thus helping avoid misinterpretation by grassroots cadres. [Chinese]
The original post at Diyin includes links to source materials, some of which have since gone offline.
One now deleted report at Nanfang Daily, archived at CDT Chinese, focused on DeepSeek’s use in online public opinion monitoring in a Shenzhen subdistrict, noting public suspicions—denied by subdistrict officials—that AI systems may be used to delete or respond to public posts by local residents:
On February 20, the Xixiang subdistrict of Shenzhen’s Bao’an district held a training session on “DeepSeek + Public Opinion Response” for 50 leaders from departments and communities within its jurisdiction. Its goal was to draw on AI technology to enhance the subdistrict’s overall public opinion monitoring, analysis, and response capabilities, and enable it to more rapidly assess and respond to the concerns and requests of city residents.
While the news received “Likes” from many netizens, others had doubts: were government departments using AI in negative ways, such as responding to or deleting online posts? Xiao Bin, propaganda chief for Xixiang’s Office of Party, Government, and Legislative Affairs, responded that AI is not being used for content deletion and other such tasks, only for rapidly sifting through and organizing vast quantities of data to ensure that every resident’s concerns and requests can be handled and answered promptly.
Grassroots Cadres: More Professional Confidence, Less Stress
At the start of the training, the instructor emphasized the necessity of learning and using AI: “You won’t be replaced by AI, but by people who know how to use it.”
They explained the main points of DeepSeek’s technical features, applicability, and operation using a threefold framework of feature analysis, application scenarios, and practical demonstration. In the public opinion response segment of the training, the instructor gave a comprehensive introduction to workflow and methodology for online public opinion response, from monitoring and analysis to response and resolution.
The instructor stressed that in the digital age, public opinion can spread faster and further, so each subdistrict department and community needs to heighten its vigilance and, putting the people first, promptly respond to the masses’ concerns. At the same time, malicious rumors and incitement must be firmly dealt with in accordance with the law. The session also included practical discussion and simulated drills.
A propaganda committee member from Xixiang’s Liutang community commented, “In the past, I’ve often felt flustered when dealing with public opinion. We never had all the information, and responses were haphazard. Now that I understand DeepSeek’s formidable capabilities, like quickly picking the crucial points of public opinion from the flood of online data and performing quantitative analysis of public opinion trends, I can be more confident in follow-up response work, and less stressed.”
An official from Xixiang described the training as an important effort to innovate the subdistrict’s grassroots governance methods, and said that by adopting advanced AI technologies, they would be able to more effectively capture public opinion data and create a positive public opinion environment amenable to the subdistrict’s harmonious and stable development.
Subdistrict Clarifies: No AI Post Deletions
Xixiang’s “DeepSeek + Public Opinion Response” training drew widespread public attention. While much of this was positive, there was also suspicion among some netizens.
Therefore, I decided to speak with Xiao Bin, the propaganda chief for Xixiang’s Office of Party, Government, and Legislative Affairs. Xiao Bin explained that the subdistrict has always attached great importance to public opinion work. Using platforms such as “Quick Response to Public Opinion,” “Mediation and Dispute Resolution,” “Bao’an District’s ‘Scan for Instant Resolution of Unpaid Wages,’” and “Bao! You’re Hired!” to quickly gather information on the public’s various complaints and comments so that “when the people call, we answer.”
As a major Shenzhen subdistrict with over a million inhabitants, Xixiang processed, verified, handled, and responded to 1,050 public opinion cases of one kind or another in 2024. Issues like market supervision, urban management, transportation, and the environment featured heavily. Handling these cases brought swift resolution and response to a great many of the problems reported by city residents.
Xiao Bin refuted claims that government departments use AI for negative activities like post deletions. She noted that a real-name account is required for replying to posts, and registering pseudonymous accounts for automatic AI-generated responses would be not only technically difficult, but also probably illegal. She said the subdistrict’s public-opinion workers strive to be objective, fair, and responsible, taking every public-opinion case seriously and engaging in active communication with the public for the sole purpose of conscientiously resolving their problems.
“The significance of AI in public opinion work,” Xiao Bin said, “is in helping us rapidly filter and compile masses of data so that relevant departments in the subdistrict can accurately assign resources to quickly verify and deal with various issues, and ensure that every resident’s concerns and requests receive prompt resolution and response.”
Asked about plans for future work, Xiao Bin said that they would continue to optimize public opinion work mechanisms, treating the public’s online views and suggestions as “opportunities to deliver government services right to people’s doors,” and taking full advantage of AI to further improve the efficiency of public-opinion handling and provide the public with even better and more effective service, promoting harmonious and stable societal development. [Chinese]
At the Journal of Democracy last month, Valentin Weber placed DeepSeek in the context of the PRC’s steadily evolving surveillance ecosystem, writing that it “has massive potential to enhance China’s already pervasive surveillance state, and it will bring the Chinese Communist Party (CCP) closer than ever to its goal of possessing an automated, autonomous, and scientific tool for repressing its people.” The shortcomings of AI outputs, such as “hallucination” of false positives, only add to the potential risks of such a system.
There is some degree of performativity in the surge of very vocal DeepSeek adoption. The New York Times’ Meaghan Tobin and Claire Fu noted that “The enthusiastic embrace of the technology by China’s bureaucracy reflects, in part, what often happens when Mr. Xi, China’s most dominant leader in decades, puts his stamp of approval on something. (Mr. Xi has set off frenzies over soccer, winter sports and high-end manufacturing, for instance.) […] But it can be hard to parse the substance from the hype. While scores of officials have pledged to use DeepSeek in their work, few have described specific examples in which the technology has made that work more effective or efficient.”
Similarly, Wired’s Zeyi Yang wrote that while some companies “have found genuine uses for the domestic, affordable AI model with cutting-edge capabilities, […] others are merely doing it for the publicity boost or to virtue-signal their national pride.” (Yang noted that this AI gold rush does not have uniquely Chinese characteristics: “The whole frenzy resembles what happened in late 2022 when ChatGPT launched and a wave of American and European companies scrambled to find ways to signal to customers and investors they were engaging with what was then the most cutting-edge innovation in AI.”)
But there’s also another factor that has helped make DeepSeek particularly trendy in China: the fact that the West freaked out about it. “Its strong reception overseas has further boosted its popularity in China, serving as the firm’s best marketing campaign,” says Angela Huyue Zhang, a law professor who studies Chinese technology policy at the University of Southern California.
The narrative that DeepSeek is challenging US dominance in AI has contributed to a growing sense of national pride within China. A central part of the company’s heroic origin story is its development of resource-efficient models, which was seen as a direct response to US policies designed to cut off China’s access to cutting-edge semiconductors. As a result, DeepSeek’s success has fueled a growing belief in China that those measures may eventually fail.
“Where there is blockade, there is breakthrough; where there is suppression, there is innovation,” Wang Yi, China’s minister of foreign affairs, said in a speech on March 7 in which he also compared DeepSeek to China’s previous technological breakthroughs in areas like nuclear weapons development and space exploration. [Source]
The performative aspects of the DeepSeek wave, and the arguably overhyped technical leap that DeepSeek represents, do not necessarily diminish its significance. The Economist argued last week that “The true winner of the AI race […] may not be the country that invents the best models. It is more likely to be the country where governments, businesses and ordinary people use AI at scale every day. For everything from economic growth to military power, technological diffusion ultimately matters more than technological innovation. On that front, the race is closer than many in America believe.”
#DeepSeeks #Outstanding #Results #Field #Public #Security #Public #Opinion #Response
