China plans upgrade of information attacks on U.S. social media with AI tools, new report argues

A version of this story appeared in the daily Threat Status newsletter from The Washington Times. Click here to receive Threat Status delivered directly to your inbox each weekday.

China’s military and the ruling Communist Party have turned to artificial intelligence to boost the impact of propaganda and influence operations through American social media platforms, according to a new study of Beijing’s covert operations.

The People’s Liberation Army (PLA), in particular, is conducting three types of operations against the U.S. where AI tools have greatly increased the impact, a report by the Rand Corp. states.

Once fearful of social media as a threat to their power, China’s leaders are now embracing the use of Facebook and X as key tools for influencing foreign public opinion, the think tank stated in a 183-page study made public this week.

The report is based on an in-depth review of Chinese party and military writings that analyze the goals of information warfare, what the People’s Liberation Army is calling the “Three Warfares” — public opinion warfare, psychological warfare and legal warfare.

The report examines the work of the man who is described as China’s leading military information warfare researcher, Li Bicheng, who specializes in cyber-enabled influence operations. Mr. Li is leading the Chinese military into adopting AI to run large-scale networks of automated bots to influence both domestic and foreign public opinion, the report contends.

Based on the writings, the report says the PLA has begun using AI and plans to expand the use of more sophisticated bots for waging “public opinion struggle,” along with propaganda bots, follower bots and “roadblock bots.”

The PLA has already been detected using the generative AI system ChatGPT with a cyber group known as Spamouflage, an online propaganda and disinformation operation identified earlier this year. PLA researchers also are using other U.S. open-source AI models, the report said.

AI will also help improve the efficiency of current Chinese campaigns called “astroturfing” — deception operations intended to promote the false impression of widespread public support for an issue.

“On the basis of cracking the enemy’s ability to verify the account information for social bots, the social bot will complete public opinion struggle and intelligence-collection tasks, such as predicting the attributes of artificial users, analyzing their social behavior network, destroying the enemy’s opinion leaders, speeding up the dissemination of the enemy’s redundant content, burying its effective information, comprehensively disrupting the enemy’s information order, intensifying its internal contradictions, and causing a strong large-scale effect,” Mr. Li is quoted in the report as saying.

Influencing elections

Chinese military researchers have studied how to use social media to influence the outcome of U.S. elections. The methods highlighted by U.S. intelligence officials were China’s efforts to influence the 2022 midterm elections by magnifying divisions in the U.S. on social media.

Mr. Li wrote articles in 2019 and 2020 that quoted a Chinese intelligence official Wang Danna, regarding “online public opinion warfare modes, responses, and prospects in the U.S. presidential elections,” including a study of former President Donald Trump’s first impeachment.

In one case, a Chinese researcher, Hu Huaping, conducted a study that analyzed the impact of social media on Mr. Trump’s impeachment.

“It is clear that Hu was specifically working on generating content that would have been highly tailored for U.S. political discourse on social media,” the report said.

Another PLA researcher writing on “social media warfare” collected X data from 30 senior officials of the Biden administration, including President Biden’s own official account and all his Cabinet secretaries.

U.S. intelligence officials told reporters last month that propagandists in China, Iran and Russia are using artificial intelligence to create content designed to deceive Americans ahead of the November presidential election.

AI is “a malign influence accelerant, not yet a revolutionary influence tool,” an intelligence official said. “In other words, information operations are the threat, and AI is an enabler. Thus far, the [intelligence community] has not seen it revolutionize such operations.”

Beijing’s social media manipulation efforts since the 2010s so far produced limited results. However, the use of new generative AI tools is expected to “dramatically improve China’s capabilities moving forward, posing a greater threat to global democracies,” the report said.

Protecting the regime

China’s military justifies its social media operations based on fears the U.S. is seeking to undermine the communist regime, the report states, noting that PLA social media manipulation capabilities began in the mid-2010s and have been underway since at least 2018.

The PLA views social media as “an acute threat to PLA prestige and regime security,” the report says.

Today, the PLA is engaged in “cutting-edge” technology, including AI, to target the American public for supporting Beijing’s strategy and policies.

The report urges the U.S. and other global democracies to prepare for AI-driven social media manipulation by adopting risk-reduction measures. Social media companies are being urged to step up efforts to combat fake and hostile accounts in order to dilute the potency of AI manipulation by China.

A greater understanding of China’s government-wide activities to manipulate social media are needed, the report states.

One organization in particular should be targeted — the United Front Work Department, a Chinese Communist Party organ with an estimated budget of several billion dollars annually.

China’s government banned Facebook, X and YouTube in 2009 in the early days of the social media revolution, after those platforms were used by dissidents to highlight mounting unrest in Tibet and western Xinjiang.

But by 2013, Chinese President Xi Jinping ordered the government to redouble propaganda efforts on foreign social media and sharply increased covert and overt influence campaigns abroad.

The operations were further strengthened beginning in 2018 as part of a return to Marxist-style “public opinion struggle” against capitalist forces that Mr. Xi believed were seeking to overthrow his government, the report said.

Military role

The PLA has played the major role in shaping the foreign influence operations through the “three warfares” strategy, which has been upgraded with technological breakthroughs of recent years.

The military now calls its efforts “intelligentized public opinion warfare,” the report said.

“This broader PLA view of the evolving nature of warfare and the desire to embrace emerging technologies is driving PLA researchers to develop a new operational concept that looks poised to replace, or at least supplement, the traditional three warfares: cognitive domain operations,” the report said.

Mr. Li, the PLA information warfare expert, shifted the focus of Beijing’s influence campaigns, saying previous efforts suffered from a lack of sophistication that made it easy for adversaries to identify Chinese online operations. AI technology is now being used to better delivery Chinese content online.

“In the field of online public opinion guidance, it is necessary to respond to this explosion of information by automatically generating a large number of posts,” Mr. Li wrote in a 2023 article explaining how to influence public discussion using large-scale activities.

China’s paid online troll operatives, known as the “50 Cent Army” for the alleged reward for each posting, also was inefficient despite 15 years of effort, the report said.

Mr. Li also said China’s online influence campaigns were less effective than those of the U.S. and, therefore, needed reform.

Artificial intelligence and so-called intelligentization operations will reduce and eliminate human labor in social media manipulation but still use human oversight, the report said.

The PLA plan for AI-driven social media operations involves a six-step process, beginning with discovery and acquisition of key information, preparation and selection of media carriers and the production of tailored content for each of the targeted online platforms.

The system then would select the appropriate timing, delivery mode, and steps for conducting the operations. Dissemination is then strengthened to create “hot spots” that will be used to form desired effects and further shape the environment and expand influence.

Mr. Li’s AI model seeks to conduct big-data mining and online opinion surveillance and to block “enemy” content that China opposes.

Then, content or “ammunition” for PLA information operations is produced based on specific target audiences, using “image- and video-stitching technology,” along with “multi-media camouflage technology.”

The operations can produce a targeted victim’s voice, specific background and specific scenes to form multimedia information with images, text and sound.

Last, the PLA will launch large-scale online information delivery using different media, including text messages, emails, forums, instant messaging tools, microblogs (such as Sina Weibo) and social media platforms, including Facebook and X.

All the operations will be organized through an “online public opinion struggle command and control center.”

Supercharged

The report said China is using U.S.-origin open-source AI technology already and could apply them to Mr. Li’s influence program.

“Available evidence suggests Beijing is using open-source models and attempting to covertly use other publicly available closed-source models,” it states. “The advent of generative AI has finally delivered the technical capability to realize Li’s early vision for an AI-driven social media manipulation system.”

Chinese officials argue the AI push is a reaction to U.S. and allied attacks on China’s own systems. Mr. Li is quoted as saying the West is engaged in an “online ideological offensive toward China” in an effort to undermine the communist system.

“This is also an attempt to dissolve our political beliefs, concept of the overall situation, and organizational discipline; to separate the party and cadres from the masses; and to deny our culture, thereby completely toppling our country,” he said.

Mr. Li has long promoted the use of social bots, the Rand researchers said, and “the PLA and likely other parts of the party-state are following in his footsteps and have expressed interest in developing such a capability.”

“Generative AI is very likely to supercharge these efforts, making this an important topic to watch.”

China’s military recently carried out a reform of its Strategic Support Force, spinning off its information warfare section into a new “PLA Cyberspace Force.” The new cyber force is expected to speed up the adoption of AI in information warfare operations, the report said.

The report, “Dr. Li Bicheng, or How China Learned to Stop Worrying and Love Social Media Manipulation: Insights into Chinese Use of Generative AI and Social Bots from the Career of a PLA Researcher,” was written by Nathan Beauchamp-Mustafaga, Kiernan Green, William Marcellino, Sale Lilly and Jackson Smith.