China’s military working on AI weapons and systems for warfighting and ‘overthrowing regimes’

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

China’s military is working on artificial intelligence weapons and technical systems that the Communist Party-ruled government in Beijing hopes will boost the People’s Liberation Army’s war-fighting capabilities.

The PLA, in recent months, has disclosed work on ChatGPT-like artificial intelligence systems in military writings, according to a report by a U.S. Air Force think tank highlighting China’s push to make its forces the first in the world to bring AI-powered systems to the battlefield.

The Air Force’s China Aerospace Studies Institute assessed in the report that the PLA is building AI weapons-related systems in seven main areas, including human-machine ties, rapid decision-making, network warfare, cognitive warfare, wartime logistics, space operations, and military training.

The report, published under the title “China’s ChatGPT War,” was made public Monday by the institute, part of the Air University located at Maxwell Air Force Base, Alabama.

ChatGPT is produced by the U.S. company OpenAI and can mimic humanlike responses to prompts with AI-generated content.

Analysts predict the technology will eventually produce text, video and images based on advanced software combined with high-speed computing that can mimic human brain creation.

China’s development of similar chatbots and technology has caused unease among U.S. national security officials.

The Biden administration recently restricted exports of advanced computer chips to China over concerns that the chips will assist China’s drive for AI-powered weapons.

Using AI for human-machine relations is the most significant force multiplier for the PLA, according to the Air Force report.

Chat-GPT-like programs will be integrated into unmanned combat systems that can rapidly collect and assess information and plan attacks on enemies.

Decision-making is a second area being worked on by the PLA to produce AI war-fighting tools that seek to increase the speed and autonomy of commanders’ war plans and attacks.

Both areas are already a focus of U.S. advanced war-fighting developments.

The Pentagon is speeding up work on a joint domain command and control system that will use AI in a bid to better deter a Chinese invasion of Taiwan.

China, meanwhile, is also working on AI-powered command and control systems.

State media in Beijing last month highlighted the recent car crash death of Feng Yanghe, a Chinese defense expert on AI-driven military systems. The Communist Party-aligned China Daily noted that Mr. Feng “had made important breakthroughs in the use of AI to support military command and control operations.”

AI coded cyberattacks

Gen. Mark A. Milley, the outgoing chairman of the Joint Chiefs of Staff, stated in a recent journal article that advanced technology, including AI, is producing seismic changes in the nature of war.

“Low-cost autonomous platforms, coupled with commercial imagery and behavior tracking data augmented by artificial intelligence and analysis tools, will accelerate the ability to sense and make sense of the environment,” Gen. Milley wrote.

“Inexpensive drones, loitering munitions, and precision-guided munitions with increasing speed, range, and accuracy will further reduce the time it takes to close the kill web.”

Both the Pentagon and China recently conducted tests of swarming drones that demonstrated AI-powered capabilities to operate in unison.

Cyberwarfare is another area of focus for China.

According to the China Aerospace Studies Institute report, PLA authors wrote recently that AI will write code and carry out offensive cyberattacks.

“Generative AI will be used by hackers to design, write and execute malicious code, build bots and websites to trick users into sharing their information and launch highly targeted social engineering scams and phishing campaigns,” the report said.

The technology also aims to produce advanced “polymorphic” malware that can rapidly evade network security software, the report said.

For PLA cognitive warfare, it said, PLA researchers believe “ChatGPT will raise the intensity of cognitive warfare to a whole new level.”

Cognitive warfare is the use of non-kinetic means to influence enemy leaders and populations in support of PLA military objectives.

Examples include using AI chatbot software to mimic a senior government official to sow confusion in adversary forces or gain access to sensitive material, the report said.

“At the societal level generative AI can ‘efficiently generate massive amounts of fake news, fake pictures, and even fake videos to confuse the public,’” it stated, quoting a PLA report.

Ultimately, the PLA plans to use AI in its cognitive war fighting to “destroy the image of [an enemy] government … divide society and overthrow the regime,” stated one PLA report from April.

China’s military separately plans to use AI to improve PLA military logistics in war fighting through planning transit routes and autonomously delivering support materials directly and accurately.

The China Aerospace Studies Institute report also assessed that the PLA aims to use AI to build and monitor space systems and enhance military training to create more realistic combat scenarios.

A key weakness of the Chinese military, which has been engaged in a decadeslong buildup of advanced weapons and forces, is a lack of war-fighting experience, the report said.

It assessed that the PLA is wary of using the self-thinking technology in combat over worries that such AI systems cannot be controlled or trusted. Chinese military officials fear that their own AI-systems could be penetrated by adversaries and used against the PLA or could facilitate efforts to undermine the ruling Chinese Communist Party.

The report notes China’s recent implementation of interim regulations for the more than 100 AI providers in China. The regulation requires all content produced by chatbots to support “socialist core values” — the euphemism for China’s communist ideology.

“If Generative AI must conform to the ‘truth’ the CCP deems as correct, certain applications and information provided could be severely tainted or compromised,” the report said.

“ChatGPT is still a black-box model … and it cannot be guaranteed that it will not generate attacks or even harm users,” the report quoted two PLA researchers as writing in April.

The PLA also worries about the “Frankenstein complex” – AI military systems achieving some self-awareness that would then produce catastrophic effects for the Chinese military, according to the China Aerospace Studies Institute report.

The report concludes, however, that the PLA regards chatbot AI technology as potentially leading to a revolution in war fighting.

“The PLA most certainly wants to be the first mover on applying a more comprehensive application of generative AI on the battlefield, but they will not do so until they can fully trust the technology,” the report said.

The report was written by Josh Baughman, an analyst with the China Aerospace Studies Institute.

‘Leapfrog’ technology

Gregory C. Allen, an AI expert with the Center for Strategic and International Studies who worked on AI for the Pentagon, said China regards military uses of AI as a high priority and is using the technology to build lethal autonomous weapons.

Mr. Allen, who spent three years at the Pentagon’s Joint Artificial Intelligence Center, has said China’s ambitious strategy for AI weapons is “moving beyond any sort of on-the-battlefield human supervision into increasingly autonomous AI-enabled warfare.”

In testimony to the congressional U.S.-China Economic and Security Commission, Mr. Allen quoted Zeng Yi, a senior official at the Chinese state-run NORINCO defense contractor as saying: “In future battlegrounds, there will be no people fighting.”

China’s military views AI as a “leapfrog” technology allowing the PLA to win against a militarily stronger United States in a conflict.

Low-cost, long-range autonomous unmanned submarines are one area.“China believes these systems will be a cheap and effective means of threatening U.S. aircraft carrier battle groups and an alternative path to projecting Chinese power at range,” Mr. Allen said.

CIA Director William Burns testified to Congress earlier this year that ChatGPT technology will be used by foreign intelligence services for “compelling spear phishing” cyberattacks.

“It’s logical to use artificial intelligence of one kind or another to produce a message that can be pretty effective in spear phishing and therefore in taking advantage of vulnerabilities,” he said.

FBI Director Christopher A. Wray said during a speech at the World Economic Forum in Switzerland in January that he is “deeply concerned” China’s AI use will be used in Beijing’s hacking operation, intellectual property theft and repression of dissidents.

Rick Fisher, a China specialist, said Beijing’s development of AI for military uses is likely well underway and will be incorporated into advanced combat aircraft, unmanned and manned hypersonic platforms, and PLA unmanned “joint force” warfare that will seek to integrate autonomous air, land and sea combat systems.

China is expected to achieve some level of AI-enabled combat capability “sooner rather than later,” he said.

“This constitutes a huge reason for the U.S. to invest in Russia- and China-levels of regional, tactical nuclear weapons to deter their conclusion that regional nuclear superiority plus AI can win wars against America,” Mr. Fisher said.