The Center for Security and Emerging Technology (CSET) employs several Chinese-capable researchers and has machine and human translation capabilities for Chinese and other languages. The following CSET product summarizes Chinese media reaction to three AI-related policy initiatives or statements made by entities associated with the U.S. government in the fall of 2019.
- The Bureau of Industrial Security of the U.S. Department of Commerce added 28 Chinese organizations to its Entity List on October 7 due to their suspected involvement in human rights abuses in China’s Xinjiang Autonomous Region. Most of these newly added entities are Xinjiang regional government bodies, but several are Chinese surveillance and security companies involved in facial recognition and other AI-related technologies. The U.S. government and U.S. companies are prohibited from doing business with entities on the Entity List (subject to some significant exceptions).
- The Defense Innovation Board (DIB) of the U.S. Department of Defense released a document entitled AI Principles: Recommendations on the Ethical Use of Artificial Intelligence by the Department of Defense on October 31.
- The National Security Commission on Artificial Intelligence (NSCAI) issued its Interim Report to Congress on the state of U.S. AI research and technology on November 4.
To gauge PRC media reaction to these developments, CSET researchers conducted Chinese-language web research using keyword searches related to the events above. We used the Google and Baidu search engines as well as the search function in popular Chinese messaging app WeChat. We conducted our research between November 14, 2019 and January 27, 2020.
PRC Media Reaction to the Addition of 28 Chinese Entities to the Entity List
The PRC government criticized the United States for interference in its domestic affairs after the U.S. announced the additions to the Entity List on October 7, which include leading Chinese AI companies iFlyTek, Megvii, Sensetime, and Yitu. The affected companies generally downplayed the immediate impact of this move on their bottom lines. PRC technology and financial commentators disagreed as to whether the long-term impact of the designations on the Chinese AI sector would be positive or negative.
The Chinese government defended its Xinjiang policy and rebuked the United States for interference in China’s internal affairs.
- PRC Foreign Ministry Spokesman Geng Shuang, at a press conference on October 8, said that China expresses “strong dissatisfaction and firm opposition” to the U.S. move, and contended that the United States was merely “manufacturing an excuse” to interfere in China’s internal affairs.
- Geng declared that “there are no human rights issues in Xinjiang,” and characterized China’s policies there as legitimate counterterrorism and counter-extremism measures that “accord with international practice” and are “universally supported” by the people of Xinjiang.
A number of the Chinese companies added to the Entity List released statements of their own dismissing the impact of this designation on their business operations.
- Liu Qingfeng, chairman of speech recognition firm iFlytek, in an “internal letter” to employees dated October 9 that later circulated widely online, claimed that iFlytek had prepared a contingency plan in case it was added to the Entity List, and stated “no power can stop us,” citing the company’s revenue growth and independent intellectual property rights.
- PRC data recovery firm Meiya Pico, in an October 9 public statement, reassured investors that it owns the rights to its own technology, that the components it buys from overseas can be “easily replaced” by domestic parts, and that its foreign sales account for less than 1% of company revenue.
- Hikvision executive Huang Fanghong, in an interview with Chinese tech news website Leiphone published January 6, said that she expected the company’s “negative growth” in the U.S. market to continue. However, Huang maintained that Hikvision’s supply chain was largely unaffected, as the company relied on “normal commercial-grade” chips and components from domestic providers.
Articles on Chinese investing and technology websites expressed mixed opinions about the long-term effects of the Entity List designations on the PRC tech sector. Commentators pointed out that the designated PRC companies were now cut off from the specialized U.S. AI chips they depended on. But these writers also speculated that this situation would spur Chinese chip manufacturers to innovate and improve.
- An article posted to the public WeChat account of Securities Times on October 9 warned that while companies such as Hikvision can find alternate suppliers for their low-end products, their more sophisticated devices, and those of China’s AI startups, rely on imports of advanced chips such as field-programmable gate arrays (FPGAs) that could be affected by the designation.
- The Securities Times post also warned that many Chinese AI companies’ algorithms are based on U.S. underlying core algorithms, and expressed concern that these too could potentially be impacted by the Entity List.
- The article concluded, however, that China has “only one way to go” now, namely developing its own advanced AI chips. Securities Times is owned by the Communist Party mouthpiece People’s Daily, but its output does not necessarily reflect authoritative party views.
- Writing in his public WeChat account on October 9, PRC technology commentator Tie Liu said that Chinese AI companies currently rely on specialized AI chips designed by U.S. manufacturer Nvidia, which are superior to domestically produced AI chips. The Entity List designation, however, would “force” Chinese firms to buy domestic chips, and after a “very painful” process of adjustment, in the end this would “help” China’s domestic AI chip industry.
PRC Media Reaction to the DIB AI Principles
The sparse PRC media commentary on the October 31 publication of the DIB AI Principles was largely negative. Chinese writers portrayed the Principles as “false ethics” because they do not disavow the use of AI-enabled lethal weapons systems. Two prominent state-run technology think tanks published articles about the Principles, suggesting that Chinese policy circles are discussing them.
- In an article that appeared in The Beijing News on November 4, Li Zheng of the Institute of American Studies at the China Institutes of Contemporary International Relations (CICIR), an influential PRC foreign policy think tank associated with China’s main intelligence agency, argued that none of the five Defense Department AI ethics principles espoused in the DIB document restrict lethal military applications of AI. Li concluded that the U.S. has “entirely abandoned” the principle of “AI demilitarization.”
- Li wrote that although the Principles “appear strict,” they are designed to coax the American public into accepting an AI-armed military. Li claimed the Principles are also inconsistent with the EU’s and China’s stances against the “militarization of AI.”
- In a similar vein, Yang Yuze, a columnist for Changjiang Daily, argued on November 4 that the Principles are “false ethics” because they “run counter to” various international statements that call for a moratorium on lethal autonomous weapons systems. The Principles “only cover the relationship between AI weapons systems and U.S. military personnel” and “give virtually no consideration” to the morality of the use of such weapons on their targets, according to Yang.
Despite their official affiliations, Li’s and Yang’s articles do not necessarily represent the consensus view of the Communist Party.
- The Beijing News, although published by the Propaganda Department of the Beijing Municipal Party Committee, is a commercially-oriented mass-market paper, not the authoritative voice of the city leadership, represented by its sister paper, Beijing Daily.
- Changjiang Daily is the official newspaper of the Wuhan City Party Committee, and articles in the paper labeled “editorial” or attributed to a “staff commentator” convey the viewpoint of the city leadership. Yang’s article appeared under his own byline, however, and thus does not officially represent the party line.
In an indication that the Principles may have caught the attention of Chinese policymakers, two influential state-affiliated technology think tanks posted lengthy Chinese-language summaries of the Principles on their public WeChat accounts. Both articles were informational and avoided making value judgments about the Principles.
- The China Academy of Electronics and Information Technology, a think tank under state-run electronic systems manufacturer China Electronics Technology Group Corporation—which in turn is subordinate to China’s ministerial-level State-Owned Assets Supervision and Administration Commission—published its summary on November 6.
- The Internet Law Research Center of the China Academy of Information and Communications Technology (CAICT) released its article on the Principles on November 15. CAICT is run by the Ministry of Industry and Information Technology.
PRC Media Reaction to the NSCAI Interim Report
Although official Communist Party media outlets were not observed to editorialize on the NSCAI Interim Report issued on November 4, commercially oriented media depicted the document as driven by the “not entirely unjustified” fear that the United States is falling behind China in AI. These articles also highlighted the importance of Chinese talent in U.S. AI innovation, and opined that any U.S. disengagement from PRC companies and researchers in the name of security would be counterproductive for the United States.
Many Chinese articles about the Interim Report pointed out flaws in the U.S. AI ecosystem.
- TMTPost, a technology news portal with a large youth readership, published a lengthy interview of NSCAI Chairman Eric Schmidt and Vice Chairman Bob Work on November 18. The TMTPost interviewer asked if Google’s 2018 withdrawal from Project Maven—a Department of Defense initiative to use AI to improve object recognition by military drones—was a harbinger of a broader trend of U.S. private sector disengagement from the Defense Department, to which Work responded in the negative.
- Claiming that the PRC government will invest $150 billion to achieve global leadership in AI by 2030, the interviewer depicted U.S. AI spending as meager in comparison. He noted that Schmidt and Work could not provide a figure for U.S. government spending on AI, and TMTPost reported, based on its own research, that the U.S. federal government budgeted less than $1 billion for non-defense AI research in fiscal year 2020.
- Duowei News, a Beijing-based news portal aimed at an overseas Chinese audience, reporting on November 8 on an NSCAI conference held the day after the Interim Report was issued, claimed that President Trump’s political support comes from “traditional industries” that do not understand AI’s potential, and that U.S. political partisanship impedes all aspects of U.S. government decision making, including AI policy.
- The Duowei article also argued that the U.S. AI industry is dominated by giant “monopolistic” companies such as Google and Facebook that divert investment and market share away from smaller, more innovative startups.
PRC media seized the occasion to play up China’s strengths as a burgeoning AI power.
- A November 8 article by Forward Business Intelligence Company, a private news and research provider based in Shenzhen, mocked U.S. officials at the NSCAI conference for being “nervous” about a Chinese kindergarten textbook on AI highlighted at the event. Despite its criticism of U.S. “alarmist rhetoric,” however, the article conceded that in the case of AI, US fears are “not entirely unjustified.”
- Rejecting the notion that the Chinese AI industry is dependent on U.S. hardware, the same article cited Dahua CEO Li Ke as claiming that his firm only imports about 10% of its components from North America, and that it has built up a large enough reserve of such “irreplaceable” parts to ride out more than a year of sanctions if need be.
- The Duowei article conceded that China’s AI chip industry “lags significantly behind” the United States, but asserted that China’s advantages are the “foresight and boldness” of its government, its huge market, its greater “big data resources,” and its larger number of data engineers.
Noting language in the NSCAI report about the danger of IP theft and economic espionage posed by a subset of Chinese researchers in U.S. universities, labs, and companies, Chinese media emphasized the negative consequences for the United States of cutting off research collaboration with China.
- A second Duowei article on November 15 argued that the White House was “overreacting” to the threat of Chinese industrial espionage, asserting that the harm to U.S. scientific research from restricting U.S. companies from doing business with China is a greater risk to U.S. national security than are attacks plotted by “overseas hostile forces.”
- The TMTPost interview of Schmidt and Work noted their opposition to complete cessation of technology and talent exchange with China, but implied that U.S. “political circles” favor an overall strategy of disengagement from China.
A policy research organization within Georgetown University, the Center for Security and Emerging Technology produces data-driven research at the intersection of security and technology, providing nonpartisan analysis to the policy community. CSET is currently focusing on the effects of progress in artificial intelligence and advanced computing. We seek to prepare a new generation of decision-makers to address the challenges and opportunities of emerging technologies.