レポート 3451

China’s spy agency has warned of the risks posed by artificial intelligence technology, especially those that could threaten national security. The Ministry of State Security said the rapid development and wide application of AI – such as ChatGPT – posed challenges that “may fundamentally change the existing national security landscape in the near future”.

In an article posted to its WeChat account on Thursday, the ministry listed data leaks, cyberattacks and data manipulation as potential risks, as well as the technology’s impact on the economy and military.

Earlier this year, State Security Minister Chen Yixin warned of cybersecurity challenges that could threaten China’s infrastructure. He also wrote in an article published in September that technologies such as AI had brought new and greater uncertainties for security.

Beijing put forward a framework for artificial intelligence at last month’s belt and road forum, calling for equal rights for all nations. The framework – known as the Global AI Governance Initiative – notes the huge opportunities of artificial intelligence but also its “unpredictable risks and complicated challenges”. AI governance is among the areas Beijing and Washington have pledged to cooperate on, according to a Chinese statement following Wednesday’s summit between Xi Jinping and Joe Biden in San Francisco. In the article, China’s spy agency warned of the risks of data leaks, noting that AI requires massive amounts of data for learning and training and that it could include a lot of sensitive information. The ministry said any misuse of information or leaks could pose a serious threat to the privacy of individuals and to national security.

It also warned of cyberattacks that the ministry said could be assisted by AI, with hackers able to “easily launch targeted and highly concealed attacks against specific targets”.

The agency also pointed to the risk of “data poisoning” – or “adding malicious data into the training data of AI”. It said this could disrupt the normal operation of data analysis models, giving the example of disruption to intelligent car technology that could lead to accidents. Another example was the targeting of propaganda campaigns, which it said could result in “negative thoughts” discreetly infiltrating public opinion.

In addition, the article raised concern over the prospect of AI technology replacing human workers in some areas, which it said could have an impact on the country’s economic, social and political security. The ministry said the technology could also be used to cause financial market panic through the manipulation of information.

It said there were risks in the use of AI in lethal autonomous weapons, and in its use to make military actions more targeted and precise by connecting networks, decision makers and operators.

The ministry said there was a need to tighten laws and regulations on the technology, as well as to promote global governance and crack down on crime involving AI.

“Addressing the risks from AI and seizing the strategic initiative of AI development to effectively safeguard national security is an important issue in global national security governance,” the article said.

It called for improved “strategic confidence” and coordination on the development and safety of AI and for a “prudent” response to the risks and challenges.