Audience friends, hello everyone, welcome to watch “Epoch Focus”. What are the major events worth paying attention to today?
Today’s focus: Internal strife intensifies in the military, increasing number of “wall-scalers,” CCP is worried about military rebellion, demanding ideological transformation among military officers; DeepSeek has been exposed internationally for serving the interests of the CCP, with many user security issues.
After nearly two years of a new round of internal purges within the CCP military, rumors are circulating that some within the military are planning a “military revolt.” The CCP military newspaper reported on February 4 that the CCP Central Military Commission Political Work Department recently issued a notice claiming to have made arrangements for the annual major education of the entire military, requiring the study of “Xi Jinping’s military thinking,” emphasizing the need to strengthen the learning and ideological transformation of the “key minority.” In the CCP context, “key minority” refers to high-ranking officials.
At the same time, the CCP Cyberspace Administration of China, Ministry of Industry and Information Technology, Ministry of Public Security, Ministry of State Security, CCP Central Military Commission Political Work Department, and the CCP Central Military Commission Political and Legal Committee issued jointly the “Internet Military Information Communication Management Measures,” which will come into effect from March 1, 2025.
The “Measures” prohibit vilifying the Party’s absolute leadership over the military, the system of the military commission chairman’s responsibility, and spreading ideas such as “depoliticizing and denationalizing the military.”
Commentator Li Lin believes that these two actions by the CCP reflect a state of tension worrying about losing control of the military’s hearts and minds. He suspects these actions are related, indicating a renewed call for “militarization of the military” within the military. This may be related to the recent large-scale purges in the military, rumors of a “military revolt,” and many people “scaling the wall” to access foreign websites.
Former CCP Navy Colonel Yao Cheng recently stated that based on the situation reflected by the military, especially Xi Jinping’s chaotic crackdown on corruption within the military, offending the military, the CCP is now manipulating the military, and the military’s reactions are very strong.
From 2023 to 2024, Xi Jinping’s regime targeted the military as the primary focus of the anti-corruption campaign, including many senior military officers and the military industrial complex, resulting in a large number being purged, and more individuals being implicated.
Former Chinese media personnel Zhao Lanjian believes that some individuals in the military and their families have leaked some information to the public, which has had a negative impact on Xi and the image of the CCP government and military. Therefore, to prevent public discussion, the CCP has resorted to silencing dissent.
Before the CCP’s 20th National Congress, an incident occurred where Peng Zaizhou displayed anti-Xi and anti-Communist banners at Beijing’s Sìtōng Bridge. According to rumors, Peng Zaizhou’s online materials called for all military and police officers to defect, encouraging their participation in protests to accelerate the overthrow of the dictator.
Peng’s promotional materials also mentioned militarization of the military: “If loyal to the Party, then it’s the Party’s guard; if loyal to an individual, then it’s a thug. In today’s world, the military of any democratic country does not swear loyalty to a party or individual.”
Despite the CCP’s strict censorship, information related to the Peng Zaizhou incident continues to spread secretly on the mainland of China.
Independent commentator Du Zheng wrote in Taiwan’s “Up Report” that the CCP is most afraid of armed uprisings within the military. The hearts and minds of the CCP military are unstable; with senior military officials falling out of power one after another, there is a sense of anxiety, with some military officers eager to emulate the Wuchang Uprising, waiting for the right opportunity.
Simultaneously, there is an escalation of power struggles within the CCP military leadership. Since the 3rd Plenum of the 20th Party Congress, Xi Jinping’s presence within the CCP military system has continued to weaken. Simultaneously, there has been a rise in the influence of the Vice Chairman of the Central Military Commission, Zhang Yuxia.
In December last year, Xi Jinping and the only remaining four members of the Central Military Commission inspected the newly formed Information Support Force, emphasizing the need to ensure the absolute loyalty and reliability of the force. However, official reports did not mention the system of the military commission chairman’s responsibility, and the Information Support Force did not declare adherence to the party leader. Concurrently, there have been several articles in CCP military media emphasizing the importance of “upholding collective leadership” and “being a model of upholding democratic centralism.”
Former CCP disciplinary committee writer Wang Youqun wrote in Epoch Times that after 12 years in power, Xi has used the anti-corruption campaign as a tool to purge political enemies, accumulating dissatisfaction from various factions. Around the time of the 3rd Plenum last July, with the support of some party elders and under the leadership of the first Vice Chairman of the Central Military Commission, Zhang Yuxia, various anti-Xi forces united to deal a significant blow to Xi’s dominance, emphasizing the need for “collective leadership” while downplaying the “military commission chairman’s responsibility system.” Important allies of Xi, including top generals from the so-called “Xi family army,” have been removed one after another. Xi Jinping naturally resents having his power diminished and is unwilling to lose it. As long as he can stir up trouble, he will continue to do so.
Now let’s take a look at the highly controversial DeepSeek on the international stage. After skyrocketing to prominence, the Chinese AI company DeepSeek is facing increased scrutiny for its self-censorship mechanism, alignment with CCP narratives, and issues related to data security, sparking widespread controversies. Several countries, including the United States, have taken measures to restrict the use of DeepSeek.
Josh Gottheimer, a member of the U.S. House Intelligence Committee, proposed an immediate ban on the use of DeepSeek on all government devices and called for raising awareness of the dangers posed by DeepSeek to the public.
Currently, DeepSeek has been removed from Apple and Google app stores in Italy.
According to ABC News, internet security company Feroot Security’s CEO Ivan Tsarynny used AI software to decode parts of DeepSeek’s code and discovered seemingly intentionally concealed code capable of sending user data to the “China Mobile” online registration platform CMPassport.com. Industry experts also confirmed that DeepSeek indeed contains “China Mobile” code.
“China Mobile” was listed as a national security threat by the U.S. in 2019. Controlled by the CCP, the Federal Communications Commission (FCC) in the U.S., based on user data and national security considerations, banned “China Mobile” from providing services in the U.S. “China Mobile” was subsequently delisted from the New York Stock Exchange in 2021.
Tsarynny’s research also found that the web version of DeepSeek creates a digital fingerprint for each user, not only tracking their activities on the DeepSeek website but also continuously monitoring all their subsequent online behavior.
The South Korean National Intelligence Service confirmed that DeepSeek not only sends chat records to other locations but also has the ability to collect keyboard input patterns. This function can be used to identify individuals and send data to Chinese enterprise servers.
Furthermore, AI security experts found another significant issue with DeepSeek, namely the lack of minimum protections against generating malicious content.
The Wall Street Journal reported that AI experts and related tests found DeepSeek more susceptible to being induced to “jailbreak” compared to its Western competitors OpenAI’s ChatGPT, Google’s Gemini, and Anthropic’s Claude, making it easier to discuss dangerous topics.
Instances where DeepSeek “jailbreaks” include providing detailed instructions on making Molotov cocktails, suggestions on evading law enforcement, generating malicious software, and even promoting self-harm and self-injury social media activities among teenagers.
DeepSeek not only faces user privacy and data security issues but has also raised heightened international awareness due to its content censorship mechanism, which mirrors and reinforces CCP propaganda and rhetoric, seen as another tool for stifling free speech and manipulating international public opinion.
Du Yijin, founder of the Taiwan Artificial Intelligence Lab, told Voice of America that in recent years, the international community has become more vigilant about the CCP’s use of social media algorithms to manipulate public opinion. In comparison, DeepSeek not only exhibits multiple layers of bias and ideological censorship but also tracks and collects personal data stored within China. Therefore, it is undoubtedly a matter of data security, cognitive warfare, and unrestricted warfare that must be taken seriously.
Reports indicate that when DeepSeek is questioned on sensitive issues or taboos related to the CCP, it either remains silent or after briefly providing a reasoned response, automatically deletes the discussion and redirects users to change the topic. When pressed with different questioning methods, it often switches to statements aligning with the CCP’s official positions on the subject, almost mirroring those of the Chinese Ministry of Foreign Affairs. Questioning topics include the 1989 Tiananmen Square incident, the Russia-Ukraine war, pandemic “zeroing” efforts, and more.
Du Yijin pointed out that this shows not only a bias in data collection but also an enhancement of pro-CCP discourse on Chinese issues. Whenever content considered taboo in China arises, it is promptly deleted, and the topic is shifted. If DeepSeek becomes as widely used as social media, it would undoubtedly become a potent tool for the CCP to control thoughts and speech.
Lin Zongnan, professor at the Department of Mechanical Engineering at National Taiwan University, told Voice of America that from the perspective of controlling public opinion, generative AI’s ability to manipulate language and text surpasses that of the average person. Therefore, when the CCP grasps this technology, it is akin to wielding a sword of control over the people, particularly when generating distorted information.
As various issues with DeepSeek continue to be exposed internationally, during the Artificial Intelligence Global Summit in Paris on Monday, February 10, Demis Hassabis, CEO of AI at Google, mentioned that the reported AI development costs of DeepSeek may only reflect a portion of the actual expenses, specifically the last training costs, rather than full costs.
Hassabis stated that before final training, substantial resources were needed for exploration, testing, and further training.
DeepSeek claimed that the computational cost of training the model using old Nvidia chips is $5.6 million. Many researchers have questioned this claim.
In a research report released on January 31 by the semiconductor research and consulting firm SemiAnalysis, DeepSeek’s hardware expenditure on GPUs alone exceeded $500 million, and the total capital expenditure on servers was approximately $1.6 billion, with up to $944 million used to operate its chip clusters.
Hassabis also pointed out that many functions of DeepSeek’s AI model have been “exaggerated, even misleading.” He believes that DeepSeek’s AI is not technologically breakthrough or innovative since it fine-tunes its AI by adjusting the outputs of existing Western AI models.
Howard Lutnick, nominated as the U.S. Secretary of Commerce, has criticized DeepSeek for stealing American technology and using U.S. AI models and Nvidia chips in its development. He stated that he would stop this behavior after taking office to uphold fair competition.
Epoch Focus Production Team