The security risk associated with Chinese AI lies in products that send user data to Chinese servers, not in the AI models themselves

Overview Debates Related claims Topics Sources

The first AI disaster will be a self-replicating AI worm, not a rogue superintelligence

Small open-weight AI models should be restricted because they could self-replicate

Global regulation of AI is difficult to achieve when the United States and China do not trust each other to conform

Good faith dialogue is needed, particularly on biosecurity, to limit the harmful use of AI

China has a history of using its hacking ability to consistently target U.S. cyber infrastructure

AI models require regulation in domains such as cybersecurity and virology, particularly in a Chinese context

US-developed AI models also have inherent risks when considered in isolation

The security risk from Chinese AI models is diminished if the US is developing similar technology at a comparable rate

A security risk from Chinese AI models is their potential application in virology and pathogen boosting

Developing an offensive AI hacking capacity does not automatically create an equivalent defensive capability

The United States and China see AI development as a necessary offensive capability

China will use AI to enhance its hacking ability, building on its history of targeting U.S. cyber infrastructure

Chinese AI models do not pose a greater security threat than American AI models

Chinese AI models, considered in isolation, have inherent risks

AI could be used to boost the virulence of pathogens, creating biosecurity risks

China is using its leadership in the AI race to compete against the transatlantic alliance, which poses a geopolitical risk

China uses its advanced technology systems to advance its foreign policy and increase its influence worldwide

Totalitarian or communist regimes use AI technology to infiltrate national security systems

China allies with Russia and uses its cybersecurity system to protect totalitarian actors

China's actions regarding AI are not necessarily more harmful or morally different than those of other countries